Unger A. - The Successful Traders Guide To Money Management (Wiley 2021)

THE SUCCESSFUL TRADER’S
GUIDE tO MONEY
MANAGEMENT
THE SUCCESSFUL
TRADER’S GUIDE
tO MONEY
MANAGEMENT
Proven Strategies, Applications, and
Management Techniques
Andrea Unger
This edition first published 2021
© 2021 John Wiley & Sons, Ltd
Registered office
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United
Kingdom
For details of our global editorial offices, for customer services and for information about how to
apply for permission to reuse the copyright material in this book please see our website at www
.wiley.com.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the
prior permission of the publisher.
Wiley publishes in a variety of print and electronic formats and by print-on-demand. Some material
included with standard print versions of this book may not be included in e-books or in
print-on-demand. If this book refers to media such as a CD or DVD that is not included in the
version you purchased, you may download this material at http:// booksupport.wiley.com. For
more information about Wiley products, visit www.wiley .com.
Designations used by companies to distinguish their products are often claimed as trademarks. All
brand names and product names used in this book are trade names, service marks, trademarks or
registered trademarks of their respective owners. The publisher is not associated with any product
or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best
efforts in preparing this book, they make no representations or warranties with respect to the
accuracy or completeness of the contents of this book and specifically disclaim any implied
warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that
the publisher is not engaged in rendering professional services and neither the publisher nor the
author shall be liable for damages arising herefrom. If professional advice or other expert assistance
is required, the services of a competent professional should be sought.
Library of Congress Cataloging-in-Publication Data is Available
ISBN 978-1-119-79880-4 (hardback)

ISBN 978-1-119-79882-8 (ePub)
ISBN 978-1-119-79881-1 (ePDF)
Cover Design: Wiley

Cover Image: © Andrea Unger
Set in 12/14pt, PerpetuaStd by SPi Global, Chennai, India
10 9 8 7 6 5 4 3 2 1
For the members of my family, who have always stood
beside me and offered their support every day, also when
taking the most difficult decisions.
CONTENTS
Foreword xi
Preface xiii
CHAPTER 1 Martingale and Anti-Martingale 1

1.1 The Right Stake 1
1.2 Martingale 2 vii
1.3 Anti-Martingale 9
1.4 More Examples 15
1.5 A Miraculous Technique? 17
1.6 Conclusions 20
CHAPTER 2 The Kelly Formula 21

2.1 Kelly and Co. 21
2.2 Conclusions 31
CHAPTER 3 A Banal Trading System 33

3.1 Analyzing a System Based on Moving
Averages 33
3.2 Applying the Kelly Formula 37
3.3 Conclusions 52
CHAPTER 4 Money Management Models 53
4.1 The Fixed Fractional Method 54
4.2 Optimal f 60
4.3 Secure f 65
4.4 Fixed Ratio 68
4.5 Percent Volatility Model 81
4.6 Levels for Changing the Number
of Contracts 91
4.7 Conclusions 92
CHAPTER 5 Refining the Techniques 94

5.1 The Importance of the Trader’s
Temperament 94
5.2 Reduced f 95
5.3 Aggressive Ratio 97
5.4 Asymmetric Ratio 99
5.5 Timid Bold Equity 100
viii 5.6 Equity Curve Trading 103
CONTENTS
5.7 z-Score 110

5.8 Conclusions 112
CHAPTER 6 The Monte Carlo Simulation 114

6.1 Using the Monte Carlo Simulation 114
6.2 Maximum Loss 135
6.3 Conclusions 139
CHAPTER 7 The Work Plan 141

7.1 Using a Work Plan 141
7.2 Conclusions 155
CHAPTER 8 Combining Forces 157

8.1 Using a Combination of Systems 157
8.2 Portfolio Money Management 168
8.3 Which Capital? 169
8.4 The Effects of Portfolio Money
Management 173
8.5 Conclusions 180
CHAPTER 9 Money Management When Trading

Stocks 181
9.1 Trading in the Stock Market 181
9.2 Conclusions 192
CHAPTER 10 Portfolio Management 193

10.1 A Portfolio Approach 195
10.2 Some Improvements to the System 208
10.3 Conclusions 214
CHAPTER 11 Discretionary Trading 215

11.1 Trading Criteria and Definition 215
11.2 An Example: Mediaset 218
11.3 Adjusting Volatility During the Trade 225 ix
CONTENTS
11.4 Trading Futures 228
11.5 Conclusions 245
CHAPTER 12 Questions and Answers 246
APPENDIX I 252
I.1 The Impact of a Trading System on
Planning 252
I.2 The Trading System 252
APPENDIX II 268
II.1 Understanding the Type of Strategy 268
APPENDIX III 278

III.1 The Advantages of Forex 278
APPENDIX IV Online Trading 282
IV.1 The Trader 282
IV.2 Trading Profits 284
IV.3 Systematic or Discretionary? 286
IV.4 Choosing the Broker 287
IV.5 Which Platform? 288
Index 291
x
CONTENTS
FOREWORD
T he second millennium began with the explosion of online trading in

Europe, as the increase in the amount of available information and
advertising of various kinds goes to show, encouraging many to try their
luck at trading on the financial markets.
A considerable number of brokers set up shop and offered their services
both for trading online and speculation on the markets, and new books are
published almost every day, written by expert traders giving a great deal of xi
advice on how to win on the markets. There are dozens of books on scalping,
more on speculation in general, and even more on trading systems, not to
mention those on technical analysis and even a few on trading psychology.
An expert trader following the continuous evolution of these publications
can’t help but notice that (in Europe) one thing that’s missing is a book that
explains a subject that’s by far the most important for he or she who wants to
make trading their profession, in other words: money management (hereinafter
referred to simply as M.m.).
In this book the author explains all the important points of how to
manage your own capital in detail in consideration of the risk and the
far-from-remote possibility that you might lose everything before you’ve
even learnt how to place consistently winning trades on the market.
The author has explained the subject in a clear and frank way, making the
book suitable for beginners and expert traders alike, and with the obstinacy
of someone who’s learned their lessons firsthand in the field he repeatedly
emphasizes the importance of applying the right money management tech-
niques. It would be a shame not to make the most of all the secrets this book
has to offer.
The author’s desire to help the reader understand that money manage-
ment isn’t the same as using a stop-loss can be found in every chapter. Also,
the various methods discussed throughout the book, which are intrinsic to
the strategies that can be applied to manage assets, let the trader prepare a
plan of action for their M.m. that’s as close as possible to perfect.
Anyone who reads this book will realise that technical analysis, trading
systems, and various methods – no matter how valid they may be – are all
but worthless without the effective management of your assets, and it would
be a real shame if the reader failed to make the effort to apply some of the
numerous suggestions they could make on their own after reading the book.
The author, however, advises against using a poor strategy with the metic-
ulous application of M.m. even if that can produce acceptable results, and
this should perhaps make us reflect on the fact that the correct application
of M.m., as well as protecting yourself from the risk of going bankrupt, can
also help you obtain spectacular results that would be impossible without
correct money management.
I love reading books on trading (I don’t think many people have a collec-
xii
tion as vast as mine on the subject), and I can truly say this book on money
FOREWORD
management is a must, and is the first complete and clear book to come out
of Italy on how to apply M.m. to financial markets. I was lucky enough to
be given the chance to read it first, and made good use of numerous sug-
gestions to manage some futures’ trading strategies, so I must compliment
the author on the excellent work he’s done in creating a book that’s a real
one of a kind – a book readers would do well to read and read again, always
keeping it on hand to use as a point of reference to dispel any doubts on the
correct way to manage the method they’re adopting.
Domenico Foti
P R E FA C E
T he trading world has changed considerably over the last two decades,
and online trading has gradually transformed the sector from specialised
to ‘DIY’, expanding to become so widespread it’s now within reach of the
investor from the comfort of his or her own home. The 1999 dot.com
bubble made investing on the stock exchange more enticing than ever, with
people dreaming of getting rich quick in a world that once mostly consisted
of Treasury bills and bonds. When the bubble burst in spring 2000, it was, xiii
to say the least, painful for most of those who’d ventured into speculating
on the stock exchange, and produced a variety of effects, leaving its mark
also on those who weren’t literally swept away by the crash.
A small number of speculators managed to adapt to the new market, peo-
ple who’d been trading before the bubble, who’d already survived various
ups and downs; and those who managed to turn what had once been a reck-
less gambler into a professional trader. These survivors, in turn, had an effect
on other survivors, leading some to take the same route, revising their trad-
ing methods, and encouraging others to try and learn more about the specific
sector in order to trade safely and emulate those who’d made their name in
trading.
Gradually, more channels were created through which you could obtain
trading information, courses were organised, conferences held, and books
written promoting a variety of trading techniques.
The motto ‘Cut your losses and let your profits run’ is on everyone’s
lips, as is ‘First, don’t lose too much’. Scalping is the technique favoured
by the masses, as all you need is a fast and reliable trading platform, and
a marked propensity for interpreting short-term market movements. But
many traders, born scalpers, gradually move away from this type of trading
to try a less frenetic but perhaps also less enticing approach, and this is where
trading systems came in, selling trading signals and courses to construct the
same systems.
Those who follow me know I trade almost exclusively with automatic
systems, as this is the approach that’s best suited to planning your trading in
detail.
The year 2008 was bad for the masses, and in time there were other spo-
radic events – such as the flash crash, the Fukushima meltdown, and the crisis
of August 2015, which created more than a few obstacles to those who make
a living or are just trying to survive on the stock exchange.
The trading industry opened its doors to the masses, trying to convince
people all they needed was just a little time and money to obtain truly unbe-
lievable results. At first the Forex market was promoted, emphasising the
notable leverage that could be used, then there was a short-lived attempt to
promote trading with options, which paved the way for CFDs, once again
emphasising the concept that with little, you could make a lot. Then came
binary options, which didn’t actually have a lot to do with trading but still
promised a road paved with gold, and last but not least cryptocurrencies,
xiv
which in a way marked the end for binary options, but we’re already waiting
PREFACE
for the next fad, all riding the wave of greed.

When one mentions money management, though, these words can be
interpreted ambiguously and cause confusion. Most people see money man-
agement as the rigorous application of a stop-loss, and a set of rules that
produce risk/reward rations close to one-third. One classic example of this
is a system that aims to make a profit three times the system stop, which
is often considered a system with a good approach to money management.
Nothing could be further from the truth!
Everything that concerns position management should be considered part
of risk management, while money management is used to study what would
be the best choice in terms of the percentage of the capital to use for a trade.
Explained in such simple terms, it seems quite a banal thing a trader might
not consider that important. My hope is to convince you, in the following
pages, that this is not the case.
When I got my hands on my first futures’ trading system report, it was
clear that the monthly profit (or annual profit, depending on how you want
to consider it) produced by the single contract analyses wasn’t enough to
live on or justify abandoning everything else to dedicate my time exclusively
to that system. The first thing I thought was: What sort of profits could
I make if I used more contracts instead of just one, and in relation to the
results, with the same number of contracts, what would change during the
negative period of the system in terms of drawdown? If a system made a
profit of €10,000/year with one contract, it would certainly be interesting
to consider using five contracts to make (or it might be better to say, try
to make) €50,000; but the same system could have a maximum drawdown
of €2,500, which with five contracts would become a loss of €12,500 at a
certain time of the year, and it’s therefore important to ask yourself if you
could withstand such a loss and continue to follow the signals of the system
without qualms.
So, calculating the number of contracts to use becomes much more
important than one might think, and is even more so when you see the
effect a correct approach to making this choice can have on the results.
Personally, I interpret money management only in terms of the science
that tells me ‘how much’ to use and not how to do so. Some prefer to call this
simply ‘position sizing’ and include risk management also as part of money
management, but in my opinion this is a somewhat strained interpretation
that does no damage, apart from causing a few misunderstandings.
xv
Everything you’ll find in this book can be used to get the best out of the
PREFACE
trading system or technique you’ve decided is best suited to your needs,
in order to maximize profits. It won’t turn a losing trading system into a
winning one, because the basic concept on which the trades are made must
be laid on solid foundations.
CHAPTER 1
Martingale and
Anti-Martingale
■ 1.1 The Right Stake
As mentioned in the introduction, money management (M.m) aims to
establish the best stake to place when opening a trade or, in general, how 1
much of your capital to use in the gamble you are about to embark on.
I think we all tend to adopt quite a simple statistical approach that encour-
ages us to hope in a positive result after one or more negative results, and
to fear repeating a success after placing a successful stake. In general, this is
why you don’t want to continue after a certain number of consecutive win-
ning trades, while after a series of losing trades you’ll be sure the next one
will be a winner.
This tendency induces us to adopt a sort of risk management that, in gen-
eral, leads us to increase the stakes after a negative period (betting on the fact
that after various losses one should statistically expect a success) and reduce
them after a positive period (for exactly the opposite reason).
In this chapter, we’ll deal with this question by moving away from the
trading environment, to enter a world we’re all in any case familiar with:
that of the coin toss.
Flipping a coin to see whether it lands heads or tails is a classic statistical
example of 50% probability, and analyzing how we manage the stakes, on
the basis of one event or another, can produce some surprising results.
The Successful Trader's Guide to Money Management: Proven Strategies, Applications, and Management
Techniques, First Edition. Andrea Unger.
© 2021 John Wiley & Sons, Ltd. Published 2021 by John Wiley & Sons, Ltd.
This isn’t trading, and the intention isn’t to compare a trading system to
betting. The purpose of this first part is simply to demonstrate what might
be the best way to manage your available capital, when ‘staking’ part of it
on an event.
If we take 100 people with €100 each, I don’t think many would come out
winning if they had to bet on a series of 100 or 1,000 coin tosses. In my opin-
ion, most would lose all their capital due to inadequate risk management.
Of the resources to download, at the link https://autc.pro/guide you’ll
find the Excel file ‘HeadOrTail.xls’ you can use to run coin-toss simulations.
This is the one I used for the various examples we’ll be taking a look at.
As I said, let’s suppose we have a capital of €100 and we’ll use it for a series
of 100 and 1,000 coin tosses, ‘heads’ wins, ‘tails’ loses. The win/loss ratio
will be different for each analysis. In other words, let’s imagine we lose €1
on every stake; the amount won, on the other hand, changes as we analyze
various examples.
Stake calculation systems are mostly based on two styles that can be
grouped together as Martingale systems and anti-Martingale systems. The
first aim to increase exposure in the case of a loss; while the second only
increase exposure after a win and decrease it in the case of a loss.
2
■ 1.2 Martingale
MARTINGALE AND ANTI-MARTINGALE
The Martingale system comes from the roulette wheel, and in practice
is based on the impossibility of an infinite series of consecutive losses.
Therefore, the concept is that the more consecutive losses there are, the
greater the probability of a win next time. On this basis, the system involves
doubling the stake after every loss. If you bet 1 on the first spin of the
wheel, you’ll bet 2 on the second if the first bet lost, and if you lose again
you’ll bet 4, then 8, and so on, and when you get a winning spin of the
wheel you’ll finally have made a profit. Note that, if you get a win on the
second spin, you’d win 2, and after losing 1 on the first spin you’d be 1 up.
If you lost also on the second spin, you’d have lost 1 + 2 = 3, so winning 4
on the third spin would again give you a profit of 1. If you lost on the third
spin, you’d have lost 1 + 2 + 4 = 7, and winning on the fourth spin would
make 8, giving you a profit of 8 – 7 = 1. As this simulation continues, we
can see that, when we finally win, we make a profit of 1, just like we would
have if we’d won on the first bet.
The above is true if you double the stake, and it’s closely related to
roulette-betting systems where one bets on red and black or odd and even
numbers. In much more general terms, all approaches that simply increase
the stake after a loss, and not just ones that double it, are called Martingale
approaches; vice versa, these approaches decrease the stake after a win.
I’d like to emphasize that most people probably have a natural inclination
to prefer a Martingale-type approach.
Now let’s take a look at the simulations. The first is based on the supposi-
tion that, a win produces a profit of €1.25 for each €1 bet, while a loss loses
the €1 bet. As mentioned above, the probability a coin toss comes down
heads is 50%, so out of 1,000 tosses it should, in theory, land 500 times
heads and 500 times tails, producing the final result:
500 ∗ 1.25 + 500 ∗ (−1) = 625 − 500 = 125
€125 at the end for every €1 bet. Obviously, this is pure theory and the
situation must be studied more carefully, as must the strategy to adopt.
As we’ve said, each gambler has €100, so let’s analyze the results of 14
gamblers using the Martingale approach, which increases the stake by a fac-
tor x after every loss. Each gambler starts with a different risk percentage
and, in particular, for the first it’s 1%, the second 2%, the third 3%, the 3
fourth 4%, the fifth 5%, the sixth 10%, then 15%, 20%, 25%, 30%, 35%,

40%, 45%, and 50%.
The first gambler with a factor x = 2 on the first spin risking 1% bets €1
euro (1% of the €100 capital is €1). If he wins, he’ll again bet 1% of the new
capital €101.25 (he won €1.25), which is €1.0125. If he lost, however, he’ll
have €99, and using a factor x = 2, he’ll double the initial risk to risk 2%, so
99 ∗ 2∕100 = €1.98
If the gambler wins, he’ll go back to staking 1%; he would have won in
the previous stake
1.98 ∗ 1.25 = €2.475
and would therefore have a capital of €101.475, of which he’ll stake
101.475 ∗ 1∕100 = €1.01475
If he lost, however, he’d have
99 − 1.98 = €97.02
At this point, he’d stake 4% (double the previous 2%) and bet
97.02 ∗ 4∕100 = €3.8808
and so on.
The second gambler will immediately stake 2% equal to €2 (2% of €100)
and then proceed using the same logic; the third would start with €3 (3% of
€100) and the last, daring or reckless, would start by betting €50 (50% of
€100).
Figure 1.1 shows the results after 100 and 1,000 coin tosses. The simula-
tion produced 53 heads and 47 tails in the first 100 tosses and a total of 467
heads and 533 tails after 1,000 tosses. Note that after 100 tosses, only the
gamblers who bet less than 10% still have available funds, while those who
started with a greater risk have used up all their capital. The gambler who
started with 5% has increased his capital tenfold to €1,051.98. Note that
the gamblers’ luck was in; in fact, the wins amounted to 53% of the total.
Continuing the game, however, after 1,000 tosses, all the gamblers have lost
every last penny, perhaps also due to an unfavourable turn of events that
brought the percentage of wins in the first 100 tosses down to 46.7%.
multiple = 2
4
Martingale (increase bet after loss)
after 100 tosses after 1000 tosses

heads 53 53% heads 467 46.7%
tails 47 tails 533
% risk ending capital gain % % risk ending capital gain %

1% 221.99 122% 1% − −100%
2% 410.97 311% 2% − −100%
3% 649.53 550% 3% − −100%
4% 887.26 787% 4% − −100%
5% 1,051.98 952% 5% − −100%
10% 70.55 −29% 10% − −100%
15% − −100% 15% − −100%
20% − −100% 20% − −100%
25% − −100% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.1 Loss 1, win 1.25 – double bet after loss. Note how in the first 100 tosses
the scenario changes drastically, passing from 5% to 10% as initial risk.
multiple = 1.5

heads 53 53% heads 467 46.7%
tails 47 tails 533

1% 145.99 46% 1% 23.67 −76%
2% 204.40 104% 2% − −100%
3% 274.90 175% 3% − −100%
4% 355.54 256% 4% − −100%
5% 442.63 343% 5% − −100%
10% 759.48 659% 10% − −100%
15% 507.48 407% 15% − −100%
20% 109.39 9% 20% − −100%
25% 2.83 −97% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.2 Loss 1, win 1.25 – multiply bet by 1.5 after loss. The final result is less
‘harsh’ than the previous case, but still isn’t encouraging.
5

In Figure 1.2 the same scenario is analyzed in the case in which the gam-
blers, instead of doubling the percentage after every loss, multiply it by 1.5,
a more ‘conservative’ approach that produces less drastic results.
The entire reasoning behind this is based on the percentages rather than on
the resulting figures in euros. One could work on the basis of a hypothesis of
starting with €1 and betting €2 in the case of a loss and then €4 after another
loss, etc. In practice, this sort of approach would produce results similar to
those shown, but with slightly more marked multiplication factors. Note, in
fact, the gambler would have bet €2 instead of the €1.98 in the percentage
example and €4 instead of €3.8808; note also that a higher multiplication
factor causes more damage for the gamblers (the results of Figure 1.2 aren’t
as bad as those of Figure 1.1), so it’s easy to see that the approach based on
absolute stakes rather than percentages would have been even worse.
Figure 1.3 shows the Martingale approach, doubling the percentage on a
new series of tosses in which, out of the first 100 as many of 57 tosses were
wins and, of the 1,000 tosses, the number of wins was just over average at
502 wins, with 498 losses.
multiple = 2

heads 57 57% heads 502 50.2%
tails 43 tails 498

1% 193.33 93% 1% − −100%
2% 180.02 80% 2% − −100%
3% 24.67 −75% 3% − −100%
4% − −100% 4% − −100%
5% − −100% 5% − −100%
10% − −100% 10% − −100%
15% − −100% 15% − −100%
20% − −100% 20% − −100%
25% − −100% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.3 Loss 1, win 1.25 – double stake after loss, a particular favourable
situation in the first 100 tosses is in any case advantageous only for those who started
6 betting low. After 1,000 tosses, the results are balanced without any advantages even for
the more conservative approaches.
Despite this, the scenario is devastating and the results speak for
themselves.
Figure 1.4 shows the same results with the stake multiplied by 1.5 instead
of 2. The scenario is certainly less drastic but can hardly be considered
encouraging. As the statistics were better than in the first case, how can we
explain such a disappointing result?
A brief study of the logic behind the dynamics of increasing the stake sheds
some light on this.
Let’s suppose we start with 1%. We’re in the following risk percentage
situation doubling our bets, as in Table 1.1.
Note that after seven consecutive losing tosses, you would have to stake
128% of your remaining capital. This is obviously impossible to do, and you
can only stake all you have (100%). After another losing toss, you’ll have
lost all your capital.
multiple = 1.5

heads 57 57% heads 502 50.2%
tails 43 tails 498

1% 152.15 52% 1% 319.03 219%
2% 219.30 119% 2% − −100%
3% 300.11 200% 3% − −100%
4% 390.28 290% 4% − −100%
5% 482.03 382% 5% − −100%
10% 501.13 401% 10% − −100%
15% − −100% 15% − −100%
20% − −100% 20% − −100%
25% − −100% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.4 Loss 1, win 1.25 – multiply bet by 1.5 after loss, even increasing the
stakes in a more conservative way still doesn’t produce results that are anything to write
home about. 7

TABLE 1.1
consecutive losing coin tosses Percentage risked on next coin toss
0 1%
1 2%
2 4%
3 8%
4 16%
5 32%
6 64%
7 128% ???
8 Capital = zero
Starting with a higher percentage speeds up this process considerably, as

Table 1.2, starting at 3%, shows.
Or even starting at 5%, as shown in Table 1.3.
TABLE 1.2
0 3%
1 6%
2 12%
3 24%
4 48%
5 96%
6 192% -> 100%
7 Capital = zero
TABLE 1.3
0 5%
1 10%
2 20%
3 40%
8 4 60%
5 120% -> 100%
6 Capital = zero
The above tables show that, starting with a 1% risk and doubling the per-
centage risked after every loss, a series of 8 consecutive losses would reduce
the capital to zero. Starting on the other hand with 3% all the capital would
be lost on the seventh consecutive loss, or the sixth if we start at 5%.
Fans of statistics can calculate the probability that a series of 100 coin
tosses comes up 6, 7, or 8 consecutive times tails; they’ll see this probability
isn’t as low as you might think. If they then continue with the analysis to
include a series of 1,000 coin tosses the probability increases again. Here,
we won’t perform an analysis of this kind as it’s quite a lengthy process,
and in my opinion the results of the simulations provide a sufficiently clear
example of the risk taken.
Note that the real problem with this approach is running out of capital,
which, when you have to stake 100% you obviously run the risk of losing
everything in the case of another loss. The same goes for playing roulette
and placing your bet on red or black. Even leaving aside the fact that the
ball might land on zero, which makes the odds worse than 50–50, a gambler
could bet by doubling their stake each time they lose, but this approach could
only be used if you had an infinite capital, with no stake limit. I must ask
myself, who, having an infinite capital, would waste their time losing on the
stock exchange or at roulette?
In the simulation we’re studying, each gambler has an initial limit of €100,
and after losing that, he’ll be out of the game for good.
We’ve seen that our gamblers didn’t have a lot of luck, and the result
should discourage anyone who’s considering using this approach in the hope
of making money with it. So do you always lose everything, or almost? Not
necessarily. Up to now, we’ve considered gamblers who, using a Martingale
approach, increased their exposure in negative periods and decreased it in
positive periods. In effect, no matter how logical it might seem from a cer-
tain point of view, this approach is totally illogical when we consider that, in
practice, he who has less risks more, and he who has more risks less, which
puts the approach in an entirely different light.
■ 1.3 Anti-Martingale
9
So, what can you do? We’ve mentioned the anti-Martingale system – in

other words an approach that decreases exposure after a loss and increases it
after a win. In practice, with this approach there’s the tendency to increase
your exposure as you make profits from winnings, and close defensively in
losing periods.
In order to analyze what could have happened with this approach, we’ll
run the simulation simply using the same investment percentages for each
stake. Some might say that in this way we aren’t decreasing the stake after a
loss and increasing it after a win, but in reality that’s exactly what we’ll do.
In fact, we are not lowering (reducing) the percentage, but the capital to
which it is applied will be smaller after a loss, so we will be betting more
or less. The gambler who bets 1% of €100, in the case of a loss will have
€99 and, placing 1% again on the next bet, will stake €0.99, which is less
than the initial €1. Vice versa, in the case of a win of €1.25 on the basis of
the above rules, we will have €101.25 and staking 1% we’ll place the next
bet of €1.0125, which is more than the initial €1 stake. Therefore, using
a fixed bet percentage you ‘follow’ the trend of your capital, with more or
less exposure as it increases or decreases.
Now let’s go back to the first simulation, the one that after 100 coin tosses
produced 53 heads (wining tosses) and 47 tails (losing tosses). This time
we’ll take 15 gamblers instead of 14, adding one who places 51% of his
capital each time. Using the Martingale system this wouldn’t make sense,
because if the gambler lost and doubled the bet percentage, he’d be imme-
diately in difficulty as he couldn’t stake 102%.
As mentioned above, all the gamblers stake the same initial percentage
each time. Figure 1.5 shows the results of this approach.
After 100 coin tosses it’s immediately clear that the overall situation
looks much better than with the Martingale approach. No one has increased
their capital tenfold, but a lot more gamblers are making a profit on their
initial capital, even those who risked 30% on each bet (with the Martingale
approach, gamblers who started with 10% were losing after 100 tosses).
Certainly, the profits of the more prudent gamblers are lower than in the
previous case, and we can see this by making a direct comparison between
Figures 1.5 and 1.1; but this doesn’t weigh in favour of the Martingale
approach which, as shown above, was devastating as the coin tossing
continued. The first 100 tosses in fact were particularly favourable, while
10
Anti-Martingale

heads 53 53% heads 467 46.7%
tails 47 tails 533

1% 120.45 20% 1% 155.97 56%
2% 143.22 43% 2% 214.56 115%
3% 168.13 68% 3% 260.48 160%
4% 194.89 95% 4% 279.24 179%
5% 223.07 123% 5% 264.42 164%
10% 363.48 263% 10% 31.59 −68%
15% 434.78 335% 15% 0.17 −100%
20% 381.47 281% 20% 0.00 −100%
25% 243.86 144% 25% 0.00 −100%
30% 112.11 12% 30% 0.00 −100%
35% 36.32 −64% 35% 0.00 −100%
40% 8.05 −92% 40% 0.00 −100%
45% 1.17 −99% 45% 0.00 −100%
50% 0.11 −100% 50% 0.00 −100%
51% 0.06 −100% 51% 0.00 −100%
FIGURE 1.5 Loss 1, win 1.25 – anti-Martingale system.

the following 900 were distinctly unfavourable and Figure 1.1 shows how
all the gamblers lost their capital. The same goes if we make a comparison
with Figure 1.2, showing less aggressive Martingale gamblers.
Taking a look at Figure 1.5, what does this show us about the results after
1,000 coin tosses? One can immediately see that, not only various gamblers
still have their capital, but they also made a profit. Gamblers who bet 3%
to 5% have practically 2.5 times their initial capital. Their Martingale col-
leagues on the other hand, have lost everything.
This is an unlucky case, but still possible. 1,000 coin tosses are, it must be
said, not many for the law of large numbers, and final percentages of heads
and tails like those in question are anything but impossible (this data, in fact,
was obtained from a real probabilistic simulation done in Excel).
Now let’s take a look at the second simulation. In the first 100 coin tosses
the success rate was as high as 57% and then, after 1,000 coin tosses there
was a much more balanced distribution with 502 wining tosses and 498 los-
ing ones.
Figure 1.3 shows the harsh results of the Martingale approach for this
series, proving that with this approach it isn’t so much the final result that
makes it good or bad (in fact, after 100 coin tosses the situation was theo-
retically better than that shown in Figures 1.1 and 1.2) but rather the dis-
tribution of the coin tosses. (As shown above, it’s the number of losing 11

consecutive coin tosses that dictates matters in this case.)
We used the same distribution of coin tosses with the anti-Martingale
approach, and Figure 1.6 shows the results. Effectively, these are the figures,
after 1,000 coin tosses, the gambler by always placing a 10% stake, instead
of the €100 of initial capital, would have €77,863.87 in his pocket! Pure
science fiction? No, the potential of mathematics.
The anti-Martingale system doesn’t always work, as the various examples
in which all capital was lost (*) in Figure 1.6, go to prove. In practice, it’s
immediately obvious that more conservative approaches make more profit
than more aggressive approaches, within certain limits. Gamblers placing
large bets won’t last long, whatever approach they’re using.
(*) N.B.: In reality the capital isn’t mathematically lost as it is using

the Martingale approach, but we can consider it to be lost when all that
remains is less than €0.01. One could actually continue indefinitely
staking smaller and smaller amounts if there wasn’t a material limit
set on said minimum stakes. (You can’t stake less than 1 euro cent as
this is the smallest denomination of our currency.)
Anti-Martingale

heads 57 57% heads 502 50.2%
tails 43 tails 498

1% 131.77 32% 1% 342.48 242%
2% 171.39 71% 2% 1,032.68 933%
3% 220.04 120% 3% 2,743.80 2644%
4% 278.90 179% 4% 6,428.39 6328%
5% 319.03 249% 5% 13,288.71 13189%
10% 887.40 787% 10% 77,863.87 77764%
15% 1,656.27 1556% 15% 20,734.87 20635%
20% 2,273.74 2174% 20% 244.14 144%
25% 2,287.15 2187% 25% 0.12 −100%
30% 1,669.05 1569% 30% 0.00 −100%
35% 868.80 769% 35% 0.00 −100%
40% 314.56 215% 40% 0.00 −100%
45% 76.44 −24% 45% 0.00 −100%
50% 11.87 −88% 50% 0.00 −100%
51% 7.70 −92% 51% 0.00 −100%
FIGURE 1.6 Loss 1, win 1.25 – anti-Martingale system, results may appear
excessively optimistic but they do reflect reality after 100 coin tosses with particularly
favourable results and after 1,000 coin tosses with balanced results.
12
Figure 1.7 shows the results in the case of a luckier series of coin tosses in
which, after 1,000 tosses, the coin came up 512 heads. In this case the final
result is even more astounding. The gambler staking 10% would have closed
with €725,163.77 compared to the initial €100.
Figure 1.8 shows the results of the Martingale approach for the same sim-
ulation.
Figures 1.9 and 1.10 show another comparison in which the result of the
final distribution is less ‘balanced,’ and the coin has come up heads just 486
times.
It’s immediately obvious that, while with the Martingale approach the
final result is closely tied to the sequence of consecutive coin tosses, with
the anti-Martingale approach on the other hand it’s the final percentages
that have the greatest influence. A simulation with 51.2% of winning tosses,
in fact, produces much better results than a simulation with 48.6% wins.
Anti-Martingale

heads 51 51% heads 512 51.2%
tails 49 tails 488

1% 115.15 15% 1% 428.78 329%
2% 130.92 31% 2% 1,617.87 1518%
3% 146.96 47% 3% 5,376.71 5277%
4% 162.91 63% 4% 15,750.08 15650%
5% 178.33 78% 5% 40,694.53 40595%
10% 232.63 133% 10% 725,163.77 725064%
15% 222.76 123% 15% 587,284.20 587184%
20% 156.25 56% 20% 21,175.82 21076%
25% 79.63 −20% 25% 31.39 −69%
30% 29.06 −71% 30% 0.00 −100%
35% 7.43 −93% 35% 0.00 −100%
40% 1.29 −99% 40% 0.00 −100%
45% 0.15 −100% 45% 0.00 −100%
50% 0.01 −100% 50% 0.00 −100%
51% 0.01 −100% 51% 0.00 −100%
FIGURE 1.7 Loss 1, win 1.25 – 1,000 coin tosses slightly unbalanced in favour of the
gamblers produces very interesting results.
13
multiple = 1.5


heads 51 51% heads 512 51.2%
tails 49 tails 488

1% 135.07 35% 1% 427.89 328%
2% 170.93 71% 2% − −100%
3% 203.27 103% 3% − −100%
4% 227.41 127% 4% − −100%
5% 239.24 139% 5% − −100%
10% 94.63 −5% 10% − −100%
15% − −100% 15% − −100%
20% − −100% 20% − −100%
25% − −100% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.8 Loss 1, win 1.25 – stake multiplied by 1.5 after a loss, with the
Martingale approach the final scenario is certainly not encouraging.
Anti-Martingale

heads 52 52% heads 486 48.6%
tails 48 tails 514

1% 117.77 18% 1% 239.04 139%
2% 136.93 37% 2% 503.50 404%
3% 157.19 57% 3% 935.17 935%
4% 178.18 78% 4% 1,532.54 1433%
5% 199.45 99% 5% 2,217.16 2117%
10% 290.78 191% 10% 2,191.67 2092%
15% 311.21 211% 15% 98.46 −2%
20% 244.14 144% 20% 0.19 −100%
25% 139.35 39% 25% 0.00 −100%
30% 57.08 −43% 30% 0.00 −100%
35% 16.42 −84% 35% 0.00 −100%
40% 3.22 −97% 40% 0.00 −100%
45% 0.41 −100% 45% 0.00 −100%
50% 0.03 −100% 50% 0.00 −100%
51% 0.02 −100% 51% 0.00 −100%
FIGURE 1.9 Loss 1, win 1.25 – anti-Martingale system on the basis of unfavourable
results after 1,000 coin tosses. With a bet of up to 10%, there are in any case notable gains.
14 multiple = 1.5


heads 52 52% heads 486 48.6%
tails 48 tails 514

1% 147.42 47% 1% 662.55 563%
2% 209.83 110% 2% 1,252.98 1153%
3% 288.68 189% 3% 403.33 303%
4% 384.25 284% 4% − −100%
5% 495.24 395% 5% − −100%
10% 1,104.60 1005% 10% − −100%
15% 1,143.78 1044% 15% − −100%
20% 514.05 414% 20% − −100%
25% 69.13 −31% 25% − −100%
30% − −100% 30% − −100%
35% − −100% 35% − −100%
40% − −100% 40% − −100%
45% − −100% 45% − −100%
50% − −100% 50% − −100%
FIGURE 1.10 Loss 1, win 1.25 – stake multiplied by 1.5 after a loss. The same 1,000
coin tosses above produce acceptable results only for gamblers who start betting using a
very conservative approach. Starting at over 3% results in the loss of all capital.
■ 1.4 More Examples
In order to demonstrate the power of an anti-Martingale approach, let’s take
a look at another statistical game. In a bag of 999 balls, 333 of the balls are
white, 333 are red, and 333 are black.
We pick a ball at random from the bag and then put it back in (so we
don’t change the total number of balls or the probability of picking one
rather than another). As there are 3 possible picks, we ran this simulation
with 99 and 999 cases (instead of 100 and 1,000) to produce figures that
can be divided by 3.
The gamblers win €3 for each €1 stake in which a white ball is picked and
lose the €1 stake if a red or black ball is picked.
On the basis of the same logic of betting a fixed percentage of your capital
(initially equal to €100), Figures 1.11 and 1.12 show the final results with
two different simulations.
Anti-Martingale gain
after 99 picks after 999 picks white = 3
red = −1
white 23 23.23% white 322 32.23% black = −1
red 38 38.38% red 339 33.93%
black 38 38.38% black 338 33.83% theoretical average gain = 3
15
theoretical average loss = −1

theoretical win/loss ratio = 3
1% 91.03 −9% 1% 1,508.74 1409%
2% 80.62 −19% 2% 16,163.74 16064% optimal percentage = 11.11%
3% 69.54 −30% 3% 124,679.90 124580%
4% 58.47 −42% 4% 701,177.65 701078%
5% 47.95 −52% 5% 2,907,939.55 2907840%
10% 12.51 −87% 10% 51,515,579.09 51515449%
15% 1.89 −98% 15% 1,503,520.76 1503421%
20% 0.17 −100% 20% 131.39 31%
25% 0.01 −100% 25% 0.00 −100%
30% 0.00 −100% 30% 0.00 −100%
35% 0.00 −100% 35% 0.00 −100%
40% 0.00 −100% 40% 0.00 −100%
45% 0.00 −100% 45% 0.00 −100%
50% 0.01 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
optimal % ending capital gain % optimal % ending capital gain %

11.11% 9.68 −90% 11.11% 39,812,137.10 39812037%
FIGURE 1.11 Loss 1, win 3 – anti-Martingale system. The first 99 picks are
unfavourable for the gamblers who have suffered significant losses. The more balanced
situation after 999 picks, however, produces astonishing results.
red = −1
white 41 41.41% white 356 35.64% black =
red 26 26.26% red 317 31.73%
1% 185.70 86% 1% 5,800.72 5701%
2% 331.04 231% 2% 232,945.70 232846% optimal percentage = 11.11%
3% 567.57 468% 3% 6,577,462.79 6577363%
4% 937.44 837% 4% 132,439,921.61 132439822%
5% 1,493.84 1394% 5% 1,926,235,429.42 1926235329%
10% 9,375.31 9275% 10% 13,880,120,010,687.70 13880120010588%
15% 28,301.19 28201% 15% 115,711,901,369,977.00 115711901369877%
20% 44,794.89 44895% 20% 2,257,228,168,658.68 2257228168559%
25% 39,189.93 39090% 25% 153,439,683.21 153439583%
30% 19,484.17 19384% 30% 43.08 −57%
35% 5,543.66 5444% 35% 0.00 −100%
40% 891.77 792% 40% 0.00 −100%
45% 78.56 −21% 45% 0.00 −100%
50% 3.59 −96% 50% 0.00 −100%
51% 1.78 −98% 51% 0.00 −100%

11.11% 14,310.07 14210% 11.11% 38,647,223,516,866.40 38647223514766%
FIGURE 1.12 Loss 1, win 3 – anti-Martingale system. In this simulation, the first 99
picks were extremely favourable, as the percentage gains clearly show. The situation is
rebalanced in the subsequent picks and remains in any case slightly unbalanced in the
16 gamblers’ favour, resulting in profits that aren’t exactly easy to interpret.
The probability of picking a white ball is obviously one-third, so you have

1 chance in 3 of winning. The payout for a win is however higher than the
damage for a loss and correct capital management produces results that are
interesting enough to put a smile on your face.
Note that, once again, the simulation for which the results are shown in
Figure 1.12 produces much better results than the simulation in Figure 1.11.
Figure 1.12 in fact shows statistically unbalanced picks with more white
balls, picked 356 times compared to 322 in Figure 1.11. This difference
doesn’t change the statistics by much, but does change the result consider-
ably, once again going to prove that this management model provides a very
positive impulse for the capital.
If we take a closer look at Figure 1.11 we’ll see that after 99 picks the
scenario wasn’t at all encouraging; just 23 white balls compared to the 33
we expected (1/3 of 99), creating significant imbalance in the system. The
situation is rebalanced in the following 900 picks, not only recovering all
our losses, for those who continued with the original money management
model, but also producing surprising results.
Figures 1.11 and 1.12 show also a percentage that’s defined ‘optimal’
calculated theoretically; we’ll come back to this point in the next chapter.
In any case, you’ll see this percentage doesn’t always actually produce the
best investment results. This is because said percentage is calculated on the
basis of the theoretical probabilities that a white ball will be picked, and
therefore in consideration of the final scenario with 333 white, 333 red, and
333 black balls (or 33 white, 33 black, and red after 99 picks) while in the
real-life scenario the results are slightly different.
■ 1.5 A Miraculous Technique?

The results we’ve seen until now will certainly have caused a few jaws to
drop, and I think very few people could hope to achieve something of the
kind without studying the subject in depth. So, we’ve finally found the
answer to all our problems and are now ready to clean out the casino, right?
Unfortunately, I have to dampen your enthusiasm. What we’ve seen so far
is only effective when applied to a winning system. In the coin toss example,
a win paid out €1.25 while a loss was just €1. With a 50% probability of
winning, the system is advantageous, and therefore whether we manage to
increase our available capital (or not) depends only on whether we adopt 17
shrewd risk management.

Let’s suppose we have a series of coin tosses in which a win was €1, the
same as a loss. Figure 1.13 shows the results that, after seeing the previous
figures, might make you turn up your nose.
What’s more, these results derive from a situation that’s unbalanced in
our favour as the coin came up 512 time heads after 1,000 coin tosses. A
different situation is shown in Figure 1.14 where the coin came up heads
‘only’ 495 times, and as you can see this is a losing system.
By way of comparison, Figure 1.15 shows the same series of coin tosses
with a win paying €1.25 and a loss of €1, which puts a smile back on
your face.
The same is true for balls picked out of a bag. If white pays 3 and you lose
1 for red and black, after 999 picks the situation will be:
3 ∗ 333 − 1 ∗ 333 − 1 ∗ 333 = 333
The system would produce balanced results if white paid out 2:
2 ∗ 333 − 1 ∗ 333 − 1 ∗ 333 = 0
Figure 1.16 shows a simulation where, in fact, white pays 2, but we can
see that the situation doesn’t look good. Figure 1.17, however, shows the
Anti-Martingale
heads 51 51% heads 512 51.2%
tails 49 tails 488

1% 101.51 2% 1% 120.93 21%
2% 102.02 2% 2% 132.32 32%
3% 101.51 2% 3% 131.00 31%
4% 100.00 0% 4% 117.34 17%
5% 97.52 −2% 5% 95.07 −5%
10% 73.85 −26% 10% 7.30 −93%
15% 43.36 −57% 15% 0.04 −100%
20% 19.48 −81% 20% 0.00 −100%
25% 6.61 −93% 25% 0.00 −100%
30% 1.66 −98% 30% 0.00 −100%
35% 0.30 −100% 35% 0.00 −100%
40% 0.04 −100% 40% 0.00 −100%
45% 0.00 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
FIGURE 1.13 Loss 1, win 1 – anti-Martingale system. Winning the same amount as
can be lost there is no longer an advantage for the gambler and there are no interesting
results in the final scenario. The less daring gamblers have made some profit after 1,000
tosses, which in any case has more wins than losses (512 compared to 488).
18
Anti-Martingale

heads 52 52% heads 495 49.5%
tails 48 tails 505

1% 103.56 4% 1% 86.07 −14%
2% 106.18 6% 2% 67.03 −33%
3% 107.79 8% 3% 47.22 −53%
4% 108.33 8% 4% 30.09 −70%
5% 107.79 8% 5% 17.34 −83%
10% 90.38 −10% 10% 0.24 −100%
15% 58.67 −41% 15% 0.00 −100%
20% 29.22 −71% 20% 0.00 −100%
25% 11.02 −89% 25% 0.00 −100%
30% 3.09 −97% 30% 0.00 −100%
35% 0.63 −99% 35% 0.00 −100%
40% 0.09 −100% 40% 0.00 −100%
45% 0.01 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
FIGURE 1.14 Loss 1, win 1 – anti-Martingale system. In this case, winning the same
amount as can be lost, the scenario after 1,000 coin tosses just a little disadvantageous for
the gambler (495 wins compared to 505 losses) shows a loss in all cases.
Anti-Martingale

heads 52 52% heads 495 49.5%
tails 48 tails 505

1% 117.77 18% 1% 292.63 193%
2% 136.93 37% 2% 754.19 654%
3% 157.19 57% 3% 1,713.32 1613%
4% 178.18 78% 4% 3,433.02 3333%
5% 199.45 99% 5% 6,070.79 5971%
10% 290.78 191% 10% 16,329.24 16229%
15% 311.21 211% 15% 1,996.17 1896%
20% 244.14 144% 20% 10.74 −89%
25% 139.35 39% 25% 0.00 −100%
30% 57.08 −43% 30% 0.00 −100%
35% 16.42 −84% 35% 0.00 −100%
40% 3.22 −97% 40% 0.00 −100%
45% 0.41 −100% 45% 0.00 −100%
50% 0.03 −100% 50% 0.00 −100%
51% 0.02 −100% 51% 0.00 −100%
FIGURE 1.15 Loss 1, win 1.25 – anti-Martingale system. The game is once again
favourable for the gamblers, paying out more for wins than is lost.
red = −1
19

red 36 36.36% red 339 33.93%
black 36 36.36% black 329 32.93% theoretical average gain 2
% risk ending capital gain % % risk ending capital gain % theoretical average loss −1
1% 81.95 −18% 1% 85.30 −15% theoretical win/loss ratio = 2
2% 65.98 −34% 2% 59.85 −40% optimal percentage = 0.00%
3% 52.19 −48% 3% 34.65 −65%
4% 40.57 −59% 4% 16.61 −83%
5% 31.00 −69% 5% 6.61 −93%
10% 6.27 −94% 10% 0.00 −100%
15% 0.84 −99% 15% 0.00 −100%
20% 0.07 −100% 20% 0.00 −100%
25% 0.00 −100% 25% 0.00 −100%
30% 0.00 −100% 30% 0.00 −100%
35% 0.00 −100% 35% 0.00 −100%
40% 0.00 −100% 40% 0.00 −100%
45% 0.00 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
0.00% 100.00 0% 0.00% 100.00 0%
FIGURE 1.16 Loss 1, win 2 – anti-Martingale system. Also, in this case, a win pays
double a loss, but as the probability of losing is twice that of winning (2 balls out of 3 are a
loss), the game isn’t favourable in a balanced series of picks.
after 99 picks after 999 picks white = 2.5
red = −1
red 36 36.36% red 339 33.93%
black 36 36.36% black 329 32.93% theoretical average gain = 2.5
theoretical win/loss ratio = 2.5
1% 96.52 −6% 1% 430.44 330%
2% 85.43 −15% 2% 1,421.27 1321% optimal percentage = 6.67%
3% 76.27 −24% 3% 3,628.44 3528%
4% 65.59 −33% 4% 7,213.77 7114%
5% 56.88 −43% 5% 11,242.25 11142%
10% 18.89 −81% 10% 3,244.97 3145%
15% 3.82 −96% 15% 4.27 −96%
20% 0.48 −100% 20% 0.00 −100%
25% 0.04 −100% 25% 0.00 −100%
30% 0.00 −100% 30% 0.00 −100%
35% 0.00 −100% 35% 0.00 −100%
40% 0.00 −100% 40% 0.00 −100%
45% 0.00 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%

6.67% 44.49 −55% 6.67% 13,930.10 13830%
20 FIGURE 1.17 Loss 1, win 2.5 – as soon as a win pays out more than the loss, the
gamblers start making an interesting profit again.
same simulation in which a win pays just a bit more: White pays 2.5 instead
of 2, the game produces balanced results again:
2.5 ∗ 333 − 1 ∗ 333 − 1 ∗ 333 = 166.5 > 0
and profits are once again seen.
■ 1.6 Conclusions
The above examples make it easy to understand why the anti-Martingale sys-
tem is the one I choose to apply to money management strategies, and the
principle on the basis of which exposure is increased only as capital increases.
This is the foundation stone for the concepts illustrated in the following
chapters of the book. I hope I’ve also convinced those of you who had a
marked preference for the Martingale system. If this isn’t so, do keep read-
ing, as the book contains some good strategies for managing your capital.
CHAPTER 2
The Kelly Formula

■ 2.1 Kelly and Co.
In the last chapter I mentioned an ‘optimal’ percentage that should, in win-
ning conditions similar to theoretical conditions, maximize profits if used
systematically. This percentage is obtained using a theoretical approach based
on the Kelly formula.
21
Kelly worked at Bell Labs and developed his studies to help the phone
company AT&T analyze disturbances in long-distance calls. The method was
first published in 1956: ‘A New Interpretation of Information Rate’. You
can find the study online, but I believe only real math buffs will be able to
appreciate the subtleties of it. Personally, I just skimmed through it, acutely
aware of how little I actually know on this particular subject.
Kelly’s study was soon used by professional gamblers to calculate the per-
centage of the gambler’s capital to risk, and maximize profits.
Kelly’s criterion considers the distribution of events in the context in
which you’ll be operating: in consideration of how many times the system is
right and how many times it wins on average compared to how many times
it loses, it calculates the best percentage to increase your capital.
In practice, Kelly managed to intertwine the statistics of the system, com-
ing up with the best way to exploit them. In his study, he considers the
probability of the system winning – in other words, how many times a win-
ning event occurred compared to the total number of events – and then
considers how the wins are paid compared to the losses. If a system based on
100 events shows 40 favourable cases and 60 losing ones, the percentage of
wins is 40%. If the gambler won 4 for every win, and lost 2 for every loss,
the ratio of wins to losses would be 4/2 = 2. This input data was used by
Kelly to calculate the best way to bet on a system of this kind.
In particular, if:
W is the probability of winning
R is the ratio of average wins to average losses
then the optimal fraction of your capital to invest, according to Kelly, is
obtained as follows:
1−W
K% = W −
R
Or, if you prefer, using the same equation:
(R + 1) ∗ W − 1
K% =
R
Let’s look at a few examples, going back to our coin toss. In this game, we
know the theoretical probability of winning is 50%, so W = 0.5. The ratio
of average wins to average losses depends on the rules of the game, though.
In the previous chapter, we saw that it isn’t advantageous to place a bet
if the payout for a win isn’t higher than the risk of a loss, as there is a 50%
22 probability of winning. Let’s see what the Kelly formula has to say in the case
of a payout of €1 for a €1 risk. In this case, R is equal to 1; in other words,
THE KELLY FORMULA
R = 1/1 because the average win is equal to 1, as is the average loss. W is

0.5, as this is intrinsic to the coin toss.
Therefore:
(1 + 1) ∗ 0.5 − 1 2 ∗ 0.5 − 1 1 − 1
K% = = = = 0...
1 1 1
In practice, according to Kelly, the best thing to do in the above condition
is . . . not risk anything, which is the same conclusion we came to previously.
If we analyze what would be the optimal fraction, according to Kelly, in
different conditions such as for example with a €1.25 payout and a €1 risk.
With 1.25, therefore:
R = 1.25∕1 = 1.25
W = 0.5
(1.25 + 1) ∗ 0.5 − 1 2.25 ∗ 0.5 − 1
K% = =
1.25 1.25
1.125 − 1 0.125
= = = 0.1, or 10%
1.25 1.25
Therefore, according to Kelly, a game with a 50% probability of winning,
an average win of €1.25, and an average loss of €1, is played in the best way
(or rather in the most profitable way) using 10% of your capital each time.
Figure 2.1 shows the results of 100 and 1,000 coin tosses with various per-
centages of capital. Note that after 100 coin tosses, the percentage that made
most profit is 30%, closely followed by 25%. The percentage the Kelly for-
mula suggested was best – in other words, 10% didn’t produce the expected
results.
If we analyze the outcome after 1,000 coin tosses, however, we’ll see that
in this case Kelly was right; in fact, the capital accumulated betting 10% is
the highest of all.
So, why is there this difference?
This is easy to explain: as mentioned above, applied to a coin toss, Kelly
refers to a perfect system in which 100 tosses will produce 50 heads and
50 tails. In the example we examined, in reality the scenario in the first 100
coin tosses was very different, coming up heads 60 times. We can check to
AntiMartingale

23
heads 60 60% heads 509 50.9%
THE KELLY FORMULA

tails 40 tails 491
1% 140.96 41% 1% 400.82 301%

2% 196.10 96% 2% 1,414.00 1314%
3% 269.25 169% 3% 4,394.07 4294%
4% 364.93 265% 4% 12,037.29 11937%
5% 488.29 388% 5% 29,088.41 28988%
10% 1,733.21 1633% 10% 3,71,283.85 371184%
15% 4,516.21 4416% 15% 2,15,379.44 215279%
20% 8,673.62 8571% 20% 5,551.12 5451%
25% 12,257.71 12158% 25% 5.86 −94%
30% 12,649.80 12550% 30% 0.00 −100%
35% 9,397.35 9297% 35% 0.00 −100%
40% 4,915.02 4815% 40% 0.00 −100%
45% 1,752.71 1653% 45% 0.00 −100%
50% 407.38 307% 50% 0.00 −100%
51% 287.56 188% 51% 0.00 −100%
FIGURE 2.1 Loss 1, win 1.25 – an extremely favourable situation for the gambler in
the first 100 tosses; then the game becomes more balanced while still producing good
payouts.
see what Kelly would say if he knew the final percentage of wins before-
hand – in other words 60% (so W = 0.6):
R = 1.25
W = 0.6
(1.25 + 1) ∗ 0.6 − 1 2.25 ∗ 0.6 − 1
K% = =
1.25 1.25
1.35 − 1 0.35
= = = 0.28 or 28%
1.25 1.25
The result is that Kelly, for a system with 60% wins that pay out €1.25
with a loss of €1, suggests betting 28% of our capital, and as can be seen in
Figure 2.1, stakes of 25 – 30% produced the best results.
I obviously make this digression only for the purpose of mathematical
proof. In a heads–tails game, you will always have a 50% possibility of win-
ning (and it’s obviously impossible to know beforehand if there will be some
series of coin tosses that produces a different result). If we take a look at the
evolution of the coin tosses in this simulation, you’ll see that after the first
100, the results became more balanced again, with 509 heads and 491 tails
24 after 1,000 tosses.
Let’s take the same simulation and have a look at the results of the bets if
THE KELLY FORMULA
the payout was €1.5 compared to €1. Figure 2.2 shows the tables of the final
result.
In this case Kelly would advise as follows:
R = 1.5
W = 0.5
(1.5 + 1) ∗ 0.5 − 1 2.5 ∗ 0.5 − 1
K% = =
1.5 1, .5
1.25 − 1 0.25
= = = 0.1667 or 16.7%
1.5 1.5
After 1,000 coin tosses, the gamblers who bet 15% – 20% obtained the
best results (also in this case the 16.7% calculated is slightly conservative as
the percentage of wins isn’t exactly 50%, but 50.9%).
In the first 100 coin tosses, the situation is again biased towards higher
percentages due to the extremely lucky series. Out of curiosity, if we run
Anti-Martingale
heads 60 60% heads 509 50.9%

tails 40 tails 491
1% 163.44 63% 1% 1,406.36 1306%

2% 262.59 163% 2% 16,832.57 16733%
3% 414.81 315% 3% 1,71,835.69 171736%
4% 644.47 544% 4% 14,99,204.38 1499104%
5% 985.04 885% 5% 1,11,99,496.57 11199397%
10% 6,479.94 6380% 10% 26,80,85,22,213.39 26808522113%
15% 29,169.05 29069% 15% 16,06,79,94,49,613.95 1606799449514%
20% 91,243.22 91143% 20% 25,96,27,04,91,054.93 2596270490955%
25% 1,99,807.79 199708% 25% 1,12,50,03,02,952.46 112500302852%
30% 3,06,197.57 306098% 30% 12,00,74.499.37 120074399%
35% 3,25,630.36 325530% 35% 2,659.55 2560%
40% 2,36,183.18 236086% 40% 0.00 −100%
45% 1,13,604.69 113505% 45% 0.00 −100%
50% 34,760.27 34660% 50% 0.00 −100%
51% 25,853.88 25754% 51% 0.00 −100%
FIGURE 2.2 Loss 1, win 1.5 – a win paid 1.5 times a loss. This really boosts the
results of Figure 2.1. 25
THE KELLY FORMULA

the calculations using W = 0.6, the result is:
(1.5 + 1) ∗ 0.6 − 1 2.5 ∗ 0.6 − 1
K% = =
1.5 1.5
1.5 − 1 0.5
= = = 0.3333, or 33.33%
1.5 1.5
In fact, percentages of 30% – 35% produce the best results.
We’ve seen cases in which the actual results were better than the theo-
retical ones and, as a consequence, a bet placed on the basis of the Kelly
criterion could make us kick ourselves for missing the chance to make the
most profit.
But, although 1,000 coin tosses seems a lot, it’s still not enough to be sure
the coin will come up exactly 50% heads and 50% tails (We’d have to toss
the coin an infinite number of times.), so there may indeed be situations in
which the final result is definitely disadvantageous. Figure 2.3 shows a case
like this. As can be seen, after the first 100 coin tosses, and after 1,000, the
percentage of wins is less than the expected 50%. This series paid out €1.25
Anti-Martingale
heads 48 48% heads 477 47.7%

tails 52 tails 523
1% 107.64 8% 1% 195.27 95%

2% 114.42 14% 2% 336.14 236%
3% 120.10 20% 3% 510.44 410%
4% 124.51 25% 4% 684.15 584%
5% 127.47 27% 5% 809.75 710%
10% 119.11 19% 10% 294.16 194%
15% 81.70 −18% 15% 4.86 −95%
20% 40.96 −59% 20% 0.00 −100%
25% 14.86 −85% 25% 0.00 −100%
30% 3.83 −96% 30% 0.00 −100%
35% 0.69 −99% 35% 0.00 −100%
40% 0.08 −100% 40% 0.00 −100%
45% 0.01 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
FIGURE 2.3 Loss 1, win 1.25 – results are below statistical balance (win % below
26 50%) and the profits aren’t interesting.
THE KELLY FORMULA
per win so, according to Kelly, we’d have bet 10% as before. It’s immediately
obvious that this choice wouldn’t have been optimal for this series, though:
despite the positive final result, we’ve made 194% compared to a potential
710% if we staked just 5% of the capital.
This consideration should be an eye-opener and help us see the risks of
this kind of approach. While it’s true that a better series than expected may
mean losing the chance to make more profits, it’s also obvious that a less
favourable series than expected could cause damage we weren’t expecting.
What, for example, can we say about the results in Figure 2.4? Our
gambler, armed with his ‘new’ Kelly formula, goes swaggering off to his
friends with a coin to flip to bet on heads or tails. After 100 coin tosses,
betting 10% produced the maximum profit possible up to that point, and
he’s more and more excited about showing off his incredible newfound
gambling skills – but after 1,000 coin tosses? I don’t think he’d have much
swagger left, having just €3 of his initial €100 left. He was unlucky and only
456 of the 1,000 coin tosses were wins, but this can indeed happen and
must be taken into consideration.
Anti-Martingale
heads 50 50% heads 456 45.6%

tails 50 tails 544
1% 112.59 13% 1% 121.81 22%

2% 125.17 25% 2% 130.94 31%
3% 137.40 37% 3% 124.28 24%
4% 148.95 49% 4% 104.20 4%
5% 159.45 59% 5% 77.20 −23%
10% 186.10 86% 10% 2.71 −91%
15% 159.45 59% 15% 0.00 −100%
20% 100.00 0% 20% 0.00 −100%
25% 45.50 −54% 25% 0.00 −100%
30% 14.79 −85% 30% 0.00 −100%
35% 3.36 −97% 35% 0.00 −100%
40% 0.52 −99% 40% 0.00 −100%
45% 0.05 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
FIGURE 2.4 Loss 1, win 1.25 – the first 100 coin tosses show a balanced result,
which, however, doesn’t remain so in the following 900 coin tosses, leading to an evident 27
and harmful imbalance after 1,000 tosses.
THE KELLY FORMULA

In order to complete the picture we began outlining in the previous
chapter, let’s also take a look at what might happen in the ball-picking game
if we place bets using the Kelly criterion.
In the ball-picking game, the probability of winning is 1 in 3, as just the
white ball wins. Let’s suppose a win pays out 3 and a loss costs 1. The Kelly
formula would give us the following percentage:
R=3
W = 0.333
(3 + 1) ∗ 0.333 − 1 4 ∗ 0.333 − 1
K% = =
3 3
1.333 − 1 0.333
= = = 0.1111 or 11.11%
3 3
The 11.11% of these results is very similar to what we found was the
‘optimal percentage’ in some cases in the previous chapter.
Now let’s take a look at how the final scenario changes after a series of
picks. The results are shown in Figure 2.5.
We can immediately see that after 99 picks, an approach based on the per-
centage calculated as optimal doesn’t produce significant results. In fact, half
the capital has been lost. After 999 picks, however, effectively using 11.11%
of our available capital to place each bet would have maximised profits.
As in the case of the coin toss, we can immediately see that the first
99 picks were worse that what we hoped for. In a theoretical scenario
(and, as mentioned several times, Kelly based his criterion on theory),
we should have picked 33 white balls, while we actually picked only 27.
This was enough to produce poor results using the percentage calculated
theoretically.

red = −1
red 36 36.36% red 339 33.93%
black 36 36.36% black 329 32.93% theoretical average gain= 3
theoretical average loss= −1
28 theoretical win/loss ratio = 3
THE KELLY FORMULA
ending
% risk capital gain % % risk ending capital gain %
1% 106.65 7% 1% 2,154.00 2055%

3% 110.88 11% 3% 356,196.22 356096%
4% 108.31 8% 4% 28,07,690.12 2807590%
5% 102.96 3% 5% 1,62,31,306.92 16231207%
10% 54.47 −46% 10% 1,41,00,87,033.47 1410086933%
15% 16.01 −84% 15% 18,39,29,384.24 183929284%
20% 2.73 −97% 20% 67,270.64 67171%
25% 0.28 −100% 25% 0.10 −100%
30% 0.02 −100% 30% 0.00 −100%
35% 0.00 −100% 35% 0.00 −100%
40% 0.00 −100% 40% 0.00 −100%
45% 0.00 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
ending
optimal % capital gain % optimal % ending capital gain %
11.11% 49.02 −51% 11.11% 1,53,05,12,294.12 1530512194%
FIGURE 2.5 Results after 99 unlucky picks and 999 balanced picks. The Kelly
percentage maximizes profits if the results of the probabilistic calculations are observed.
Again, out of curiosity, let’s have a look at the calculations if we’d known in
advance we’d have picked only 27 wins out of 99 – in other words, 27.27%;
therefore, W = 0.2727:
R=3
W = 0.2727
(3 + 1) ∗ 0.2727 − 1 4 ∗ 0.2727 − 1
K% = =
3 3
1.0909 − 1 0.0909
= = = 0.0303 or 3.03%
3 3
Once again, the mathematical calculations run with hindsight confirm the
results of the bets. In fact, the gambler who bet on that series of picks at 3%
would be in the best shape after the first unlucky 99 picks.
But the scenario is quite different after 999 picks, in which the white
ball was picked as predicted by the probabilistic calculation, producing 331
whites compared to the expected 333. In this case, the percentage calculated
using the Kelly formula produces the best increment for the initial capital.
We should emphasize that the 999 picks started with the 99 initial unlucky
ones, so after 99 picks the gambler who started staking 11.11% would have
just €49.02 remaining; but continuing to stake the same percentage, trusting
29
in the statistics, after 999 picks his capital (hard to read, as it’s too large a
THE KELLY FORMULA

number) is shown in the second part of the figure.
Figure 2.6 shows another series of picks in which a win paid out €2.5 for
every €1 stake (in the previous case the payout was €3). The Kelly formula
would therefore be based on R = 2.5, while W obviously remains 0.333.
R = 2.5
W = 0.333
(2.5 + 1) ∗ 0.333 − 1 3.5 ∗ 0.333 − 1
K% = =
2.5 2.5
1.1667 − 1 0.1667
= = = 0.0667 or 6.67%
2.5 2.5
Note, in this case, that the first 99 picks were luckier than they should
have been, in theory; and, in fact, a gambler who started staking a higher risk
percentage would have made more than a gambler who stuck to the Kelly
percentage. In the scenario after 999 picks however, now quite balanced, the
gambler who started and continued staking the percentage calculated using
the formula made most profit.
after 99 picks after 999 picks white = 2.5
white 37 37.37% white 338 33.83% red = −1
red 33 33.33% red 333 33.33% black = −1
black 29 29.29% black 328 32.83% theoretical average gain = 2.5
theoretical win/loss ratio = 2.5
ending
1% 132.37 32% 1% 548.95 449%

3% 213.17 113% 3% 7,450.36 7350%
4% 259.79 160% 4% 18,707.37 18607%
5% 308.48 208% 5% 36,716.03 36616%
10% 504.60 405% 10% 32,350.66 32251%
15% 468.37 368% 15% 123.66 24%
20% 257.07 157% 20% 0.00 −100%
25% 85.20 −15% 25% 0.00 −100%
30% 17.12 −83% 30% 0.00 −100%
35% 2.06 −98% 35% 0.00 −100%
40% 0.15 −100% 40% 0.00 −100%
45% 0.01 −100% 45% 0.00 −100%
50% 0.00 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
30 ending
THE KELLY FORMULA
6.67% 416.21 316% 6.67% 66,423.88 66324%
FIGURE 2.6 A balanced scenario after 999 picks confirms the Kelly formula
The same picks are shown in Figure 2.7, in which a win pays out €3 how-
ever for a €1 loss, and the Kelly percentage is once again 11.11% as it was
calculated previously.
We’ve seen how a ‘mathematical’ approach, based on the Kelly formula,
can produce notable results; but it can also be less productive if the actual
results of the game aren’t close to the theoretical ones.
While, on the one hand, the main aim of a gambler is to maximize his
profits, it’s also obvious that, with any approach, there should be a brake
that’s applied to make staying in the game more probable.
A gambling approach based on the Kelly formula has, in time, been
applied to blackjack and other games based on probability (in fact, the rules
of some games have been changed in time also to thwart gamblers using a
systematic approach). Interesting studies have been conducted to compare
white 37 37.37% white 338 33.83% red = −1
red 33 33.33% red 333 33.33% black = −1
ending
1% 158.49 58% 1% 2,843.45 2743%
3% 355.96 256% 3% 8,05,884.22 805784%
4% 506.00 406% 4% 82,59,936.37 8259836%
5% 695.68 596% 5% 6,18,26,282.71 61826186%
10% 2,153.68 2054% 10% 18,49,91,51,926.64 18499151827%
15% 3,342.02 3242% 15% 7,73,20,49,978.62 7732049879%
20% 2,799.68 2700% 20% 86,10,642.12 8610542%
25% 1,322.11 1222% 25% 36.51 −63%
30% 358.97 259% 30% 0.00 −100%
35% 56.03 −44% 35% 0.00 −100%
40% 4.93 −95% 40% 0.00 −100%
45% 0.24 −100% 45% 0.00 −100%
50% 0.01 −100% 50% 0.00 −100%
51% 0.00 −100% 51% 0.00 −100%
ending
11.11% 2,826.88 2727% 11.11% 26,15,02,37,400.28 26150237300% 31
THE KELLY FORMULA

FIGURE 2.7 As in the previous case, the first 99 picks make a greater profit for the
more aggressive gambler than the Kelly formula, but a more balanced situation after 999
picks makes Kelly a winner.
the probability of making a profit with the safety of the approach used,
and the conclusions showed investments made using half the percentage
calculated with the Kelly formula were best. What’s more, if we consider
the results obtained up to now, it’s immediately obvious that a more
conservative approach is safer.
■ 2.2 Conclusions
Kelly, in his studies, came up with a way to use mathematics to calculate
the best risk percentage in a betting game in which the theoretical probabil-
ity of winning and the payout of said wins are known. The results obtained
applying this method are sometimes astonishing, and prove the power of a
well-thought-out approach when dealing with random phenomena.
However, theory and practice, as everyone knows, don’t always go
hand-in-hand, and expectations based on calculated probability aren’t
always (or it might be better to say aren’t ever) precisely as expected.
There will be cases in which the results are better than expected, but also,
unfortunately, cases in which they are worse. In order to protect ourselves
from unfortunate events, it’s good practice not to slavishly follow rules
dictated by mathematics, but apply limits to risk using the values obtained
from theoretical calculations as points of reference.
The above was applied to typical games based strictly on the calculation
of probability, but in the world of trading we of course have to operate in the
field we know best, and we’ll see how the things we’ve considered to this
point can be applied or used to devise a strategy that produces good results.
■ REFERENCE
Kelly, J.L. Jr. ‘A New Interpretation of Information Rate.’ Bell System Tech-
nical Journal (July 1956), 917–926. https://archive.org/details/bstj35-
4-917.
32
THE KELLY FORMULA
CHAPTER 3
A Banal Trading
System
In order to take the right path, we have to analyze concrete data and try to
apply the theories we’ll develop to the same. To do so we’ll use the results of
a trading system that’s somewhat banal, but well-suited to our lucubrations.
33
■ 3.1 Analyzing a System Based on

Moving Averages
Here we’ll analyze the results of a system based on a crossover of mov-
ing averages. This system uses 15-minute bar charts for the euro/dollar
exchanged on Forex. The results are in USD. The value of each tick is 12.5
USD, and there is 20 USD of commission and slippage on each trade.
The system uses a faster moving average and a slower moving average and
buys when the faster moving average moves higher than the slower moving
average, selling when the slower moving average is once again higher than
the faster as shown in Figure 3.1 with a 1,250 USD stop-loss per contract.
To limit operations to times not subject to the afternoon frenzy, we
decided to consider just the exchange rates from midnight to midday, as if
the market were closed in the afternoon.
FIGURE 3.1 Example of the dynamics of the system used.
In this sense, the chart was created by eliminating the data after 12.00, as
if the stock exchange closed and opened again next morning. This decision
34 was taken deliberately in order to easily obtain some losing trades that were
much worse than the max. loss planned. In fact, using a 1,250 USD stop-loss
A BANAL TRADING SYSTEM
may make you think you’re protected within that limit, but deleting half
the day’s trading, where most of the big movements occur, makes it highly
probable that you might open next day with a significant gap.
The instrument used provides good response with 25 periods for the
faster moving average and 40 periods for the slower moving average.
In order to further limit the number of system entry points, another filter
was used that only allows a long entry if the day’s high hasn’t exceeded the
previous day’s high (of course, the high recorded from midnight to midday, as
we don’t know the prices outside this range). Vice versa, a short entry is only
allowed when the day’s low is still higher than the previous day’s low. The
effectiveness of the system is based on the tendencies found in the market in
question, if the moving averages cross after previous day’s high or previous
day’s low have already been exceeded the signal is often not very valid. This
is why the above filter is used.
For your information, and so you can duplicate the system if you want,
here’s the EasyLanguage code (remember to use the times 12 a.m.–12 p.m.
in the chart settings to exclude afternoon bars):
input: slow(40), fast(25),MyLoss(1250);

if average(c,fast) crosses above average(c,slow) and highd(0)<highd(1)
then buy next bar at market;
if average(c,fast) crosses below average(c,slow) and lowd(0)>lowd(1)
then sellshort next bar at market;
setstopcontract;
setstoploss(MyLoss);
As you can see, the system is extremely banal but, as mentioned above, it
suits our purpose well.
The author’s database starts from the year 2000 and the system was tested
from that date to the end of 2005 for six years.
Some information from the TradeStation report of said system, follows 35
(Figures 3.2–3.5).

As you can see, the strategy pays off, although the results aren’t exactly
exceptional. The maximum drawdown is in fact about one quarter of the net
profit, and for many traders this wouldn’t be tenable. Apart from a difficult
first year (and perhaps many would have stopped using the strategy), the sys-
tem paid off in quite a consistent way. Note that, as expected, the maximum
loss suffered by the system is much greater than the 1,250 USD stop-loss.
These results are for a single contract, and in a certain sense, might be
considered interesting as they are; but our intention is to go further and see
what would happen if we applied some of the principles we’ve studied.
I would like to emphasise that this system wasn’t built following the classic
precepts, as the decision was taken to eliminate the most frenetic part of the
market after midday. The purpose of this strategy is to show the effects of
money management on a series of trades, rather than provide specific ideas
for trading.
36
FIGURE 3.2 Strategy report.

FIGURE 3.3 Annual strategy results.
37
■ 3.2 Applying the Kelly Formula

In the previous chapter, the Kelly formula gave us an idea of what might be
the optimal fraction of our capital to invest. So how can we apply this to a
trading system?
The data we have available is shown in Figure 3.2. ‘Percent profitable’
shows how many trades (as a percentage) were winning, W in the Kelly for-
mula (expressing the value as 0... rather than a percentage).
‘Ratio avg win / avg loss’ is the ratio of average wins to average losses, R
in the Kelly formula.
The Kelly formula:
(R + 1) ∗ W − 1
K% =
R
In which, for the case in question:
W = 0.4907 (49.07%)
R = 1.3248
38
FIGURE 3.4 Some trades with the strategy.
Using the above data, one obtains:

(1.3248 + 1) ∗ 0.4907 − 1
K% = = 0.106265 or 10.6265%
1.3248
This means that to maximize profits, we should use 10.6265% of our
capital for each trade.
This isn’t difficult to do when trading stocks: in general, you pay for what
you buy. In this case, though, we’re trading derivatives, and the value of what
you invest is dictated by the margin required for each single derivative.
FIGURE 3.5 Strategy equity curve.
39
To draw an analogy with a euro/dollar contract on CME, let’s suppose

the margin is 2,500 USD, which means that for every 2,500 dollars available,
you could work with one euro/dollar contract.
So, how to proceed? We calculate the amount of our available capital to
invest by applying the percentage obtained using the Kelly formula, and then
divide this by the margin, to find out how many contracts we can use.
Let’s look at a few examples before returning to our data on this spe-
cific case.
For example, if we have 100,000 dollars and a Kelly percentage of 25%,
the capital to invest would be 25% of 100,000, which is 25,000 USD. If the
margin for a contract is 2,500 USD, with 25,000 USD ‘investible’, we can
use 10 contracts (25,000/2,500 = 10).
If the Kelly percentage is 20%, we would have 20,000 USD to invest so,
with a 2,500 USD margin we can use 8 contracts (20,000/2,500 = 8).
These calculations are usually rounded down, as a more conservative
approach limits risks.
If the Kelly percentage was 17%, the resulting calculation for 100,000
USD would give you 17,000 to invest, which can’t be divided by 2,500. In
fact 17,000/2,500 = 6.8 contracts. In this case, even though 6.8 is closer to
7 than 6, we’d trade using 6 contracts (we’ll take a look further on at what
might happen if we rounded off in a different way).
So, to calculate the number of contracts, we’ll use the following formula:
contracts = ENT(capital ∗ Kelly ∕ margin)
where ENT is the ‘entire part of.’
Let’s go back to our example, assuming we started with 50,000 USD. For
the first trade, therefore, we have:
contracts = ENT(𝟓𝟎,𝟎𝟎𝟎 ∗ 𝟎.𝟏𝟎𝟔𝟐𝟔𝟓 ∕ 𝟐,𝟓𝟎𝟎) = ENT(𝟐.𝟏𝟐𝟓𝟑) = 𝟐
So we’d have used two contracts instead of one.
After every trade, our available capital will have increased in the case of
a winning result or decreased in the case of a loss, so we recalculate the
number of contracts and apply the result to the next trade.
In our example, the first trade made a 642.5 USD profit, so with two
contracts we’d have made 1,285 USD and, starting with 50,000 USD, we
would now have 51,285; so the following calculation is:
40
contracts = ENT(𝟓𝟏,𝟐𝟖𝟓 ∗ 𝟎.𝟏𝟎𝟔𝟐𝟔𝟓 ∕ 𝟐,𝟓𝟎𝟎) = ENT(𝟐.𝟏𝟕𝟗𝟗) = 𝟐
So we’d continue to trade with two contracts, and so on.

Applying this formula to all 964 trades made over the six-year period
produces the situation shown as a graph in Figure 3.6.
The thinner line shows the trend of the capital if trading with just one
single contract. You start with 50,000 USD and end up with 138,082.5 USD.
(In fact, Figure 3.2 shows a profit from the strategy of 88,082.5 USD, which,
when added to the initial 50,000, produces this figure.)
The thicker line shows the trend of the capital trading with a number of
contracts calculated each time, using the Kelly formula.
As you can see, for quite a long time (around half the length of time for
which the strategy was adopted), the performance obtained using the Kelly
formula was worse than what you would have made without adopting any
particular money management stratagems. Then, suddenly, profits rocket to
close with a capital of 433,090 USD six years later.
Why does the Kelly formula produce worse results than a single contract
strategy for such a long time?
600,000.00
433,090.00
400,000.00
primitive Kelly
200,000.00
138,685.00
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.6 Comparison between performance with one contract and the Kelly formula.
TABLE 3.1
Capital with Contract Capital with
Trade number Result 1 contract using Kelly Kelly contracts
1 642.5 50,642.50 2 51,285.00

2 –1720 48,922.50 2 47,845.00
3 32.5 48,890.00 2 47,780.00
4 92.5 48,982.50 2 47,965.00
5 –857.5 48,125.00 2 46,250.00
6 480 48,605.00 1 46,730.00
7 –1507.5 47,097.50 1 45,222.50
8 –1270 45,827.50 1 43,952.50
9 –2157.5 43,670.00 1 41,795.00
10 1592.5 45,262.50 1 43,387.50
We can find the answer to this question by simply taking a look at the
results of the first trades and what happened in terms of equity, as shown in
Table 3.1.
Note that, as may seem obvious, using more contracts amplifies profits
42 in the case of winning trades but also amplifies losses in the case of losing
trades; and, in our system, after the second trade the results with the Kelly
formula look worse than with just one contract.

The percentage we calculated was for the entire timeline of the system
that, sooner or later, would have borne fruit as expected and as shown by
the upswing around halfway through the period of the system.
Note also that the line produced by the Kelly percentage is actually a
pessimistic representation of the real performance that could have been
obtained from the theoretical line. As mentioned above, in fact, the number
of contracts is always rounded down, so you’re effectively using a lower
figure than the Kelly percentage. If we could use fractions of contracts, the
scenario would be quite different. Figure 3.7 shows a comparison with a
purely theoretical line applying the Kelly formula, without rounding down.
Note that if we could use fractions of contracts we would have made a
total of 807,419.77 USD final profit, which is just under double what was
obtained in practice.
This result may give you cause to reflect on why you should round the
number of contracts down, and encourage you to round the number up, or
perhaps round off to the nearest whole number.
1,000,000.00
800,000.00 807,419.77
Kelly Real Kelly

600,000.00
433,090.00
400,000.00
200,000.00
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.7 A comparison between the performance that can be obtained in the field using the Kelly formula and purely theoretical results
(thin broken line).
Figure 3.8 provides the answer to this by comparing the exact Kelly
line with a rounded-up number of contracts, and the number of contracts
rounded off to the nearest whole number (1.7 contracts becomes 2, as
does 2.1).
At a glance you won’t see much difference between the results of the three
methods and, in effect, the final capital doesn’t change a lot:
807,419.77 USD without rounding off
798,135 USD rounding up
836,130 USD rounding off to the nearest whole number
It would appear that rounding off to the nearest whole number gives the
best results, but this is a purely academic exercise: we’ll study this subject
in greater depth later and see that it isn’t prudent to take this sort of risk.
Applying the Kelly formula to trading proves interesting and the results
shown without a doubt encourage us to use it.
There are, however, various considerations to bear in mind in terms of
the theoretical results shown.
First, if we compare the chart in Figure 3.5 with any of the lines produced
by the Kelly formula, it’s easy to see that not only is the capital amplified,
44 but the fluctuations of the same are too.
In Figures 3.6 and 3.7 the line from Figure 3.5 is shown as the ‘primi-
tive.’ You’ll immediately notice this line is definitely flatter and more regular
compared to the ‘roller coaster’ line of the Kelly formula.
If we follow the line produced when we apply the Kelly formula to the
letter (therefore rounding down the number of contracts), Figure 3.9 reveals
some interesting points.
There is a first significant peak at 434,040 USD followed by a bad period
that brought the capital down to 175,172.5 USD. Later, the system produced
a new high at 450,957.5 USD, followed by a low of 301,112.5 USD.
Frankly, I have serious doubts that anyone could sit back and watch peri-
ods like this without panicking or suffering some kind of emotional conse-
quences. The trader, in fact, doesn’t know that, after that low there’ll be
other highs, and even the greatest faith in a system has its limits.
Finally, Figure 3.10 shows the most significant peak values of the line
when the number of contracts obtained using the Kelly formula was rounded
off to the nearest whole number.
1,000,000.00
836,130.00 Kelly rounded off

800,000.00 807,419.77 Real Kelly
798,135.00 Kelly rounded up
600,000.00
Real Kelly
Kelly rounded up
Kelly rounded off to nearest whole number
400,000.00
200,000.00
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.8 Comparison of different rounding off methods for the number of contracts calculated using the Kelly formula
Significative equity curve values obtained applying Kelly's formula
600,000.00
450,957.50
434,040.00
433,090.00
400,000.00
Kelly
301,112.50
200,000.00
175,172.50
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.9 Peak equity values applying the Kelly formula.

Kelly rounded off to nearest whole number
1,000,000.00
859,550.00
837,850.00 836,130.00
800,000.00
600,000.00
569,070.00
400,000.00
326,420.00
200,000.00
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.10 A half-million-dollar drawdown.

k
In this case, the final results are, as mentioned above, much more inter-
esting, but there’s a significant drawdown too: from over 800,000 dollars
the line drops to just over 300,000 dollars. I think many traders might have
been discouraged in a case such as this.
We made an absolutely reasonable hypothesis to take the right path using
the strategy proposed with an available capital of 50,000 USD, and the results
are those shown above.
Now what would have happen if we had just 30,000 USD?
Figure 3.11 shows the resulting equity lines: the primitive again with
just one contract (average thickness unbroken line), Kelly (thin broken line)
rounding down the number of contracts from the Kelly formula, and the
thick broken line obtained by rounding the values of the formula off to the
nearest whole number.
The line produced by the simple Kelly method looks strange, and after
an all-but-invisible start got bogged down at 23,500 USD, while the equity
without money management added the usual 88,082.5 USD to the initial
capital (30,000 + 88,082.5 = 118,082.5) and applying the Kelly formula
rounded off to the nearest whole number increased the capital from 30,000
to 421,245 USD.
48 No problem with the primitive, although the flat trend of the Kelly line
is perplexing. So, what happened?
The accumulated losses reduced the capital to a value that, using the Kelly
formula, produced a number of contracts less than 1 and, after rounding
down, this means using 0 contracts.
In particular, with the available data:
K% = 0.106265
Margin = 2,500
So, with the remaining 23,500 USD:
contracts = ENT(𝟐𝟑,𝟓𝟎𝟎 ∗ 𝟎.𝟏𝟎𝟔𝟐𝟔𝟓 ∕ 𝟐,𝟓𝟎𝟎)

= ENT(𝟐,𝟒𝟗𝟕, 𝟐𝟐𝟕𝟓 ∕ 𝟐,𝟓𝟎𝟎) = ENT(𝟎.𝟗𝟗𝟖𝟖𝟗𝟏) = 𝟎 ! !
With 0 contracts, obviously the system won’t be trading.

As we can see from the chart, rounding off to the nearest whole number,
which in this case would be 1, produces very different results, and the system
continues trading with 1 contract.
Starting capital 30,000 USD
600,000.00
400,000.00 421,245.00
Kelly
Primitive
Kelly arrotondato
intero più vicino
200,000.00
118,082.50
23,500.00
0.00
1 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951
FIGURE 3.11 The system with a smaller available capital.

So, is rounded off to the nearest whole number the best approach? Not at
all; also in this case, you might find the number to round off is less than 0.5,
which would give you 0.
In particular, with initial capital of 25,000 USD compared to the 30,000
USD in the previous example, we’d obtain the following equity line
(Figure 3.12).
Just a little less capital blocks the thin line at a value that’s quite close to
that of the previous case, but the concept remains valid, and this time we
apply also (and more painfully) the principle of rounding off to the nearest
whole number. An unlucky start reduces the capital from 25,000 USD to
11,205 USD, and as a consequence:
Contracts = Rounded off (𝟏𝟏 , 𝟐𝟎𝟓 ∗ 𝟎.𝟏𝟎𝟔𝟐𝟔𝟓 ∕ 𝟐,𝟓𝟎𝟎)
= Rounded off (𝟏 , 𝟏𝟗𝟎.𝟔𝟗𝟗𝟑 ∕ 𝟐,𝟓𝟎𝟎)
= Rounded off (𝟎.𝟒𝟕𝟔𝟐𝟖) = 𝟎
Simply because we can’t apply the Kelly formula in a strictly mathematical
way, but have to round off to use a whole number of contracts, we found
ourselves in evident difficulty, and had to stop using a system that would
eventually have made a profit.
This isn’t the biggest problem, though. It’s well known that the famous
50
American trader Larry Williams used the Kelly formula in the World Cup
Trading championship in which he increased his available capital from 10,000

USD to over 1 million USD in one year. It was an extraordinary result, but
Larry Williams soon stopped using this method after a number of crushing
defeats.
The main reason is that the formula is based on the past history of the
system and calculates the percentage with reference to trades from a certain
period.
Future trades may be better than those of the past, but also worse, and
putting your trust in an estimate of how a system works can be extremely
hazardous.
Let’s not forget that in trading the maximum possible loss is never defined,
and in this example the trading was done deliberately in a way to make the
stop-loss uncertain, so we cannot measure a decent average loss used in the
Kelly formula calculations. In our system, for example, we set a stop-loss
of 1,250 USD, but in Figure 3.4 there are various trades that lose more.
Why? If you take a closer look at the figure, you’ll note that these trades
all closed at 00.15 – in other words, on the first bar of the day. Evidently,
the market opened with a gap so that the stop-loss wasn’t triggered at the
Initial capital 25,000 USD
200,000.00
113,082.50
Kelly
Primitive
23,125.00
11,205.00
0.00
1 41 81 121 161 201 241 281 321 361 401 441 481 521 561 601 641 681 721 761 801 841 881 921 961
FIGURE 3.12 5,000 USD less is enough to make the technique that produced a result of over 400,000 USD, fail.
desired level. Cases of this kind are considered to be quite normal in this
line of work, and you can never put blind faith in a maximum theoretical
loss. Even leaving aside the possibility of opening with a gap, and using just
intraday techniques, there’s always the risk of losing more than you planned
as the technology used to send the orders to the market can cause delays or
even be unexpectedly blocked.
■ 3.3 Conclusions
In the previous chapter we saw how the Kelly percentage maximizes profits,
which are therefore worse if you use higher percentages. So, Kelly repre-
sents a case of limit risk, and trading at your limit when you don’t fully
comprehend the conditions you are trading in is never recommendable.
What’s more, it’s important to emphasise that, when you start using a
system, you don’t know its characteristics. Now, after recording data for
six years, we could start to use this approach and apply the Kelly formula
with the data obtained from the report. This, however, wouldn’t have been
possible six years ago, and in this case we would have expected a certain
number of trades (around 20) and, on the basis of the statistics of these, we
52 would have calculated the first Kelly formula. Then, as we went, we would
have adjusted the values by adding new information from the new trades. It’s
obvious that this approach loses a lot in terms of precision and is detrimental
to the basic idea.
For those who, on the other hand, consider the Kelly percentage to be
too aggressive, some have suggested one could use fractions of the Kelly
percentage; but at this point it’s better to study other methods to quantify
the most suitable number of contracts to trade with. In the next chapter
we’ll take a look at some of the most widely used methods and consider the
pros and cons of the same, obviously also on the basis of what difference they
might make when applied to our system.
CHAPTER 4
Money
Management
Models
In this chapter I’ll illustrate some of the most popular money management 53
methods. The approach we’ve studied up to now using the Kelly formula
attempts to adapt to the characteristics of the system, adjusting the amounts
of the bets on the basis of purely statistical data on the same. We’ve already
discussed the limits of this method and the possible negative repercussions.
When studying different approaches, the tendency is to concentrate on what
the consequences of what we decide to do might be. The methods we’ll take
a look at are based on concepts the aim of which is to consider negative
cases and limit their impact. Kelly tried to maximize profits, without wor-
rying about individual transitory events (in fact he focused on how these
combined, analyzing the final statistics). Many methods have, however, been
developed to attempt to consider the possible effect of every choice made,
step-by-step. So now there’s more focus on the effects of losses and the
impact these can have both in terms of global equity, but also psychologically.
Some of the methods we’ll take a look at start with the most negative case
possible, when the system suffers the worst loss, and on the basis of said data,
attempt to adjust the amounts used to enter trades and limit possible negative
impact.
These methods have been refined and initiated to consider also com-
binations of negative events, which generally occur in periods of marked
drawdowns. Therefore, considering what might be a difficult period to get
through for many traders, methods have been devised that attempt to get
you out of said period in the best possible way.
We’ll start by presenting the systems that aim to adjust the size of the
trade to limit loss of capital to a certain value. We’ll look at several variations
that, over the years, have been proposed by people who’ve studied these
techniques, and other methods that aim to limit fluctuations in the capital by,
within certain limits, regulating appreciation and, above all, depreciation.
■ 4.1 The Fixed Fractional Method

The fixed fractional method or fixed fraction or %f is by far the most widely
used money management method in the world of trading. There may be
different facets, but it’s always based on the same principle.
In practice, the system bases its calculations on establishing what percent-
age of your capital to risk on each trade. This percentage is referred to in
54 fact as a fraction or %f or simply f.
MONEY MANAGEMENT MODELS
The method doesn’t consider the parameters of the trading system, as

is the case with the Kelly formula; but attempts to focus on the investor’s
psychology, with the investor choosing how much they are willing to lose.
The basic method considers the system stop-loss as the loss on a single
contract. Once the available capital is known, you establish the fraction f
you intend to risk, to define the total amount of money you are willing to
lose, then this figure is divided by the stop-loss in order to calculate how
many contracts you can use.
An example is the best way to explain this principle.
Let’s suppose you have capital of €100,000 and are willing to risk 5%
of it.
Our system, for example, has a €1,250 stop-loss. Risking 5% of €100,000
means you’re willing to lose a total of €5,000. If, in theory, you could lose
€1,250 on each contract, you could use four contracts (5,000/1,250 = 4).
If, on the other hand, you want to limit the risk to 4% of the capital, that
means a potential loss of €4,000, so, you’d only be able to use three contracts
after rounding down the result of the division (4,000/1,250 = 3.2).
TABLE 4.1
Trade no. Result
1 −500
2 1300
3 700
4 −1250
5 −1000
6 2500
7 5350
8 −1250
9 600
10 350
With this method you recalculate the number of contracts after every
trade, and proceed as before.
Let’s suppose we have the series of trades shown in Table 4.1.
Now, suppose you want to limit your risk to 5%. Referring to Table 4.2,
you’ll see the various steps for building the equity with an initial capital of
55
€100,000 and a €1,250 stop-loss.

Figure 4.1 shows various scenarios for different risk percentages.
Note how the results are quite different after the tenth trade, and also how
increasing the risk percentage doesn’t necessarily increase the final result; at
least not in all cases, as can be seen from the result with a 50% risk, which
is lower than the result with a 25% risk.
After the seventh trade, a 50% risk made more than any of the others,
but on the eighth trade this was drastically put back into perspective after, in
fact, 50% of the capital was lost. A €1,250 stop-loss, which was considered
the worst hypothesis, had in fact been triggered, and in this case the trader
was willing to risk half his capital.
The fact that increasing the risk doesn’t always increase profits, appears
after all to be quite obvious if we consider that, for a trader willing to risk
100%, it would only take one losing trade at the maximum level to lose all
his capital and, therefore, be unable to continue trading.
It must be emphasized that the drawdown is not directly linked to the
percentage risked, but this, in fact, depends on the sequences of trades. The
only certainty is that by risking a percentage f, your percentage drawdown
TABLE 4.2
Trade No. Capital Amount risked Contracts Total result
−2,000 (4 contracts with €500
1 100000 5,000 (5% of 100,000) 4 (5,000/1,250)
loss each)
98,000 (€100,000 – 2,000
3 (4,900/1,250 = 3.92 to 3,900 (3 contracts with €1,300
2 lost in 1st trade) 4,900 (5% of 98,000)
be rounded down) gain each)
101,900 (€98,000 + 3,900 4 (5,095/1,250 = 4.076 to 2,800 (4 contracts with €700
3 5,095 (5% of 101,900)
profit) be rounded down) gain each)
104,700 (€101,900 + 2,800 4 (5,235/1,250 = 4.188 to −5,000 (4 contracts with €1,250
4 5,235 (5% of 104,700)
profit) be rounded down) loss each)
99,700 (€104,700 + 5,000 3 (4,985/1,250 = 3.998 to −3,000 (3 contracts with
5 4,985 (5% of 99,700)
lost) be rounded down) €1,000 loss each)
7,500 (3 contracts with €2,500
3 (4,835/1,250 = 3.868 to
6 96,700 (€99,700 + 3,000 lost) 4,835 (5% of 96,700) gain each)
be rounded down)
104,200 (€96,700 + 7,500 4 (5,210/1,250 = 4.168 to 21,400 (4 contracts with
7 5,210 (5% of 104,200)
just made) be rounded down) €5,350 gain)
125,600 (€104,200 + 21,400 5 (6,280/1,250 = 5.024 to −6,250 (5 contracts with €1,250
8 6,280 (5% of 125,600)
just made) be rounded down) loss each)
119,350 (€125,600 + 6,250 4 (5,967.5/1,250 = 4.774 to 2,400 (4 contracts with €600
9 5,967.5 (5% of 119,350)
lost) be rounded down) gain each)
121,750 (€119,350 + 2,400 4 (6,087.5/1,250 = 4.87 to 1,400 (4 contracts with €350
10 6,087.5 (5% of 121,750)
just made) be rounded down) gain each)
trade no. Result trend with various risk % starting with €100,000 using a €1,250 stop-loss
2.50% 5% 7.50% 10%
contracts equity contracts equity contracts equity contracts equity
1 −500 2 99000 4 98000 6 97000 8 96000

2 1300 1 100300 3 101900 5 103500 7 105100
3 700 2 101700 4 104700 6 107700 8 110700
4 −1250 2 99200 4 99700 6 100200 8 100700
5 −1000 1 98200 3 96700 6 94200 8 92700
6 2500 1 100700 3 104200 5 106700 7 110200
7 5350 2 111400 4 125600 6 138800 8 153000
8 −1250 2 108900 5 119350 8 128800 12 138000
9 600 2 110100 4 121750 7 133000 11 144600
10 350 2 110800 4 123150 7 135450 11 148450
15% 20% 25% 50%

trade no. Result contracts equity contracts equity contracts equity contracts equity
1 −500 12 94000 16 92000 20 90000 40 80000

2 1300 11 108300 14 110200 18 113400 32 121600
3 700 12 116700 17 122100 22 128800 48 155200
4 −1250 14 99200 19 98350 25 97550 62 77700
5 −1000 11 88200 15 83350 19 78550 31 46700
6 2500 10 113200 13 115850 15 116050 18 91700
7 5350 13 182750 18 212150 23 239100 36 284300
8 −1250 21 156500 33 170900 47 180350 113 143050
9 600 18 167300 27 187100 36 201950 57 177250
10 350 20 174300 29 197250 40 215950 70 201750
57
Equity trend for different risk percentages.

FIGURE 4.1
will be at least f. Actually, even this isn’t an absolute certainty; in fact a

percentage f is lost if the trade loses exactly the amount you decided on
when setting the stop-loss. But there may also be the fortunate hypothesis in
which none of the trades is stopped out, and the only losing trades are trades
with a slight loss, the trade being reversed.
If, for example, you are long with an open loss of €500 and there is a
short signal, you would close the trade long with a loss of €500, less than
the €1,250 set as the max. loss in the system.
While on one hand a losing trade can certainly be considered to be a mea-
sure of the psychological stress involved in constantly following the rules of
the system, the percentage drawdown is probably even more so; in other
words, in percentage terms how much your equity has dropped after reach-
ing the relevant level.
In the above example, the situation of the drawdowns on the basis of the
risk is quite different, and it’s worth analyzing it, always bearing in mind that
you never know how an investment choice will end (if we did know everyone
would choose to risk 25%, which after the 10 trades shown produced the
best result).
Figure 4.2 shows the drawdown level (shown as DD) for each calculated
equity, and the relevant percentage (DD%) for each trade.
For a better understanding of the figure, let’s analyze the first equity
figure, with a 2.5% risk percentage.
As mentioned above, the first trade lost €1,000, so after starting with
€100,000 we now have €99,000 with a drawdown of €1,000. In percentage
terms, this is exactly 1% of €100,000.
The following trade brought the equity up to €100,300, the high reached
to that point. As this is a new high, the drawdown is reset: it will be recal-
culated on the basis of this level.
Then, a new winning trade further increases the level of the equity to
€101,700, which is again an absolute high and a new starting point for cal-
culating the drawdown.
equity equity equity equity

f = 2.5% DD DD% f = 5% DD DD% f = 7.5% DD DD% f = 10% DD DD%
58 99000 1000 1.00% 98000 2000 2.00% 97000 3000 3.00% 96000 4000 4.00%
100300 0 0.00% 101900 0 0.00% 103500 0 0.00% 105100 0 0.00%
101700 0 0.00% 104700 0 0.00% 107700 0 0.00% 110700 0 0.00%

99200 2500 2.46% 99700 5000 4.78% 100200 7500 6.96% 100700 10000 9.03%
98200 3500 3.44% 96700 8000 7.64% 94200 13500 12.53% 92700 18000 16.26%
100700 1000 0.98% 104200 500 0.48% 106700 1000 0.93% 110200 500 0.45%
111400 0 0.00% 125600 0 0.00% 138800 0 0.00% 153000 0 0.00%
108900 2500 2.24% 119350 6250 4.98% 128800 10000 7.20% 138000 15000 9.80%
110100 1300 1.17% 121750 3850 3.07% 133000 5800 4.18% 144600 8400 5.49%
110800 600 0.54% 123150 2450 1.95% 135450 3350 2.41% 148450 4550 2.97%
equity equity equity equity

f = 15% DD DD% f = 20% DD DD% f = 25% DD DD% f = 50% DD DD%
94000 6000 6.00% 92000 8000 8.00% 90000 10000 10.00% 80000 20000 20.00%
108300 3100 2.78% 110200 15400 12.26% 113400 25400 18.30% 121600 31400 20.52%
116700 0 0.00% 122100 3500 2.79% 128800 10000 7.20% 155200 0 0.00%
99200 17500 15.00% 98350 27250 21.70% 97550 41250 29.72% 77700 77500 49.94%
88200 28500 24.42% 83350 42250 33.64% 78550 60250 43.41% 46700 108500 69.91%
113200 3500 3.00% 115850 9750 7.76% 116050 22750 16.39% 91700 63500 40.91%
182750 0 0.00% 212150 0 0.00% 239100 0 0.00% 284300 0 0.00%
156500 26250 14.36% 170900 41250 19.44% 180350 58750 24.57% 143050 141250 49.68%
167300 15450 8.45% 187100 25050 11.81% 201950 37150 15.54% 177250 107050 37.65%
174300 8450 4.62% 197250 14900 7.02% 215950 23150 9.68% 201750 82550 29.04%
FIGURE 4.2 Equities and the relevant drawdown levels.

The following trade, however, reduces the capital by €2,500 from the
high reached to that point of €101,700 to €99,200. The drawdown in this
case is, in fact, the difference between the maximum equity and the current
level, in other words:
101,700 − 99,200 = 2,500.
In percentage terms, 2,500 of 101,700 is 2.46%.
Another losing trade decreases the equity again to €98,200. This value
differs €3,500 from the high of €101,700 and in percentage terms €3,500
of 101,700 is 3.44%.
A winning trade will increase the equity again to €100,700, but this value
isn’t a new high and remains exactly €1,000 (0.98% of €101,700) below the
best high, up to this point.
Another winning trade increases the equity to €111,400, which is the
highest value it has yet reached, so a new starting point for the drawdown
values.
Immediately, a losing trade decreases the capital to €108,900, €2,500
less than the previous high it just reached. The drawdown at this moment is
exactly €2,500, which, as a percentage of €111,400 (we must now use this
new high as a reference), is 2.24%.
59
Finally, two winning trades increase the equity first to €110,100 and then

to €110,800. Neither of these values is a new high, and for each, the differ-
ence from the absolute maximum is calculated as, respectively, €1,300 and
€600, 1.17% and 0.54% of the maximum level of the equity.
It’s evident that, in percentage terms, the worst situation was reached
with a 3.44% loss on the high and this value is therefore highlighted.
The same logic was applied to all the other scenarios, and the value at the
moment of the greatest loss in percentage terms on the high was highlighted
for all.
Note that the maximum percentage drawdown doesn’t necessarily coin-
cide with the maximum absolute drawdown. In particular, this can be seen
in the last table in which the dreadful drawdown in percentage terms of
69.91% corresponds to an absolute value of €108,500 while, subsequently,
there is a drop of as much as €141,250; which at that time, however, was
just 49.68% of the maximum equity reached up to that point.
In general, the percentage values are considered, assuming resistance to
losses will increase as the capital increases. In other words, losing €1,000
if you have €10,000 isn’t the same thing, in psychological terms, as losing
€1,000 if your capital is €100,000. The same amount in the first cases is 10%
of the capital, while in the second case it’s just 1%, so is considered much
more tolerable.
It’s evident that greater exposure will produce very different trends, both
in terms of profit and loss. The table shows that choosing a 25% risk percent-
age would have produced the most capital after 10 trades, but one might ask
how many speculators would have been able to withstand a 43.41% draw-
down, momentarily finding themselves (but, not knowing the future, they
certainly couldn’t imagine the subsequent recovery) with €78,550 of the
initial €100,000.
There’s no one percentage that’s better than others because each leads to
subjective considerations that should always be made honestly. Personally,
I don’t think I’d be happy in any of the cases with a DD% of over 25%,
regardless of the large profits I was hoping for.
In any case, the fixed fractional method is one of the methods I prefer and
use to trade. I’ll go into this later on in the book to develop other examples
and considerations.
60 ■ 4.2 Optimal f
At this point, it’s worth taking a look at optimal f, which you may have
already heard of. I’ll try to explain it, both in theoretical and practical terms,
to clear up any doubts.
The concept was developed and explained by the mathematician Ralph
Vince, and his friend Larry Williams. Vince thought it was insane to trade
without trying to maximize profits and tried to come up with a system to
do just that. He stopped Larry Williams using the Kelly formula, consider-
ing it unsuitable for diversified systems like those used in trading where the
concept of gain/loss is characterised by various facets.
To calculate optimal f, Vince considered the trades the system had made,
and of these the worst loss, then parametrised all the trades in relation to
the result. Therefore, if the worst loss of a system was €1,200, another loss
of €600 is 0.5 (half the max. loss), while a gain of €1,800 is 1.5 (1.5 times
the max. loss, with a minus sign as the cash flow is going in the opposite
direction).
Ralph Vince presented his theory in the book Portfolio Management Formula:
Mathematical Trading Methods for the Futures, Options and Stock Markets. As he’s
a professor of mathematics who learned to trade in the field, the terms he
uses can seem confusing and be somewhat difficult to interpret. Let’s take a
look at the formula he proposes and see how to use it.
These are the terms used by Vince:
Pn – the profit of the nth trade

HPRn (holding period return of n) – the capital multiplier asso-
ciated with the nth trade
WCS (worst-case scenario) – the maximum loss in the analyzed
period
TWR (terminal wealth relative) – the initial capital multiplier
after a series of trades
f – the fraction of the capital risked on each trade
The following formulas are used:

HPRn = 1 − f ∗ (Pn ∕WCS) 61
TWR = HPR1 ∗ HPR2 ∗ HPR3 ∗ ... HPRn

Optimal f is the value of f that produces max. TWR.
In my opinion, considering it in the above way just creates confusion, so
let’s take a detailed look at a few examples.
First and foremost, let me explain what a capital multiplier is: let’s sup-
pose that in a trade you lose 10% of your capital – you’ll be left with 90%.
That trade reduced the capital by 10%, leaving 90%, so the capital multiplier
is 0.9.
Vice versa, a trade that produced a 5% gain would increase the capital by
5 percentage points and the equivalent would be a capital multiplier of 1.05.
HPR is therefore a number which, when multiplied by the capital, pro-
duces the capital after the trade of reference.
If the first trade loses 10% and the second makes a 5% profit on the
remaining capital, the final capital is obtained by multiplying the initial cap-
ital first by 0.9 and then by 1.05.
The same result is obtained if the winning trade came before the losing
one, multiplying first by 1.05 and then by 0.9 to obtain the same result.
capital
before
trade no. Result trade trade % trade HPR
1 −500 100000 −0.50% 0.9950

2 1300 99500 1.31% 1.0131
3 700 100800 0.69% 1.0069
4 −1250 101500 −1.23% 0.9877
5 −1000 100250 −1.00% 0.9900
6 2500 99250 2.52% 1.0252
7 5350 101750 5.26% 1.0526
8 −1250 107100 −1.17% 0.9883
9 600 105850 0.57% 1.0057
10 350 106450 0.33% 1.0033
final capital 106800 TWR = 1.0680
FIGURE 4.3 HPR and TWR for the series of 10 trades with just one contract.
The product of all these number gives you the final capital.
Let’s take a numerical example and have a look at the 10 trades we used
in the previous examples.
At the moment, we’ll limit the study to the results that would be
obtained using just one contract. The first trade produced a €500 loss,
which is 0.5% of the initial €100,000, and as a consequence, the HPR is
62
0.9950 (1 – 0.5/100). The second trade produced a €1,300 profit, which
is 1.31% of the remaining capital of €99,500 and the HPR is 1.0131. The
following trade made a €700 profit on the available €100,800 (99,500 +
the 1,300 profit), which is 0.69% and the HPR = 1.0069 and so on up to
the tenth trade that made a €350 profit on the available capital of €106,450,
which is 0.33% and the HPR = 1.0033.
The final capital is €106,800. Multiplying all the values in the HPR col-
umn produces the number referred to as TWR (1.0680), which one can see
is the multiplier of the initial €100,000, to obtain the final €106,800.
Applying money management principles, the result of the various trades
changes as the number of contracts used changes.
So, let’s see how we can explain Vince’s HPRn formula. In the equation,
WCS is our –1,250, as this is the worst trade of the series. With this trade the
loss is equal, in percentage terms, to the value of f chosen for our method.
This is true by definition as we decided that, in the case of a €1,250 loss (our
stop-loss) we didn’t want to lose more than f % of our capital.
So in this case the HPR is (1 – f ). In the hypothesis that f is equal to 0.1,
or chosen because we don’t want to lose more than 10% of our capital if the
€1,250 stop-loss per contract is triggered, the HPR would be 0.9 with an
available capital of 90% remaining.
If the loss was half of €1,250, obviously we would have lost half also in
percentage terms, in other words f/2, and the HPR would therefore be
(1 – f/2). In the case of a winning trade, for example equal to €1,250, it’s
easy to see that, instead of losing a fraction f of our capital, we gain it, and
the HPR is (1 + f ).
The meaning of the following formula is clear in consideration of the
above:
HPRn = 1 − f ∗ (Pn ∕WCS)
where Pn /WCS represents the ratio of the result of the trade to the worst
trade. If Pn is half the WCS this would be 0.5 and HPR = 1 – f *0.5. If we
make a €1,250 profit instead of the same loss, the result of trade Pn would
be –WCS (the opposite of the worst loss) and the Pn /WCS ratio would be
–WCS/WCS = –1 so HPR = 1 – f *(–1) = 1+ f, which is the same result
we obtained above using logic.
So, let’s calculate the various HPR and TWR for each risk percentage in
the same series of 10 trades.
An attentive observer will note that the TWR values obtained, when
multiplied by the initial €100,000, do not produce exactly the capital in the 63

examples in Figure 4.2. In the case of the 50% risk, the total was €201,750,
while the TWR 2.0726 in Figure 4.4 would produce a final capital of
€207,260. The same is true, if we take another example, for the 2.5% risk
for which Figure 4.2 shows a final capital of €110,800, while the TWR of
1.1366 in Figure 4.4 would produce a final capital of €113,660. If you’ve
been following closely from the start, you might have already come up with
an explanation for this difference. In practice, the TWR results calculated
in Figure 4.4 are purely mathematical, while the capital obtained in the
calculations in Figure 4.2 reflect the real-world limits and are subject to
adjustments due to rounding off the number of contracts to a lower whole
number, as fractions of contracts obviously can’t be used on the markets. If
the real-world profit is more than the theoretical application, this is merely
because in a losing trade there were fewer contracts than the theoretical
number.
The table in Figure 4.4 represents the process used by Ralph Vince. In
practice, once you’ve done the calculations he proposes, you need a lot of
patience, or a suitably programmed calculator, to calculate the TWR for f
HPR for %f =
trade no. Result Pn/WCS 2.50% 5.00% 7.50% 10.00% 15.00% 20.00% 25.00% 50.00%
1 −500 0.4 0.99 0.98 0.97 0.96 0.94 0.92 0.9 0.8
2 1300 −1.04 1.026 1.052 1.078 1.104 1.156 1.208 1.26 1.52
3 700 −0.56 1.014 1.028 1.042 1.056 1.084 1.112 1.14 1.28
4 −1250 1 0.975 0.95 0.925 0.9 0.85 0.8 0.75 0.5
5 −1000 0.8 0.98 0.96 0.94 0.92 0.88 0.84 0.8 0.6
6 2500 −2 1.05 1.1 1.15 1.2 1.3 1.4 1.5 2
7 5350 −4.28 1.107 1.214 1.321 1.428 1.642 1.856 2.07 3.14
8 −1250 1 0.975 0.95 0.925 0.9 0.85 0.8 0.75 0.5
9 600 −0.48 1.012 1.024 1.036 1.048 1.072 1.096 1.12 1.24
10 350 −0.28 1.007 1.014 1.021 1.028 1.042 1.056 1.07 1.14
WCS = −1250 TWR 1.1366 1.2732 1.4082 1.5397 1.7857 1.9980 2.1647 2.0726
FIGURE 4.4 HPR values for each trade and TWR for each chosen f.
that changes from 0.01 to 0.99 increasing by 0.01 (or even by 0.001). Of
all those TWRs there will be one that’s highest, and Ralph Vince called the
value of f that produced that TWR, Optimal f.
As long as we’re dealing with just a few trades, as in our example, you
can do the calculation using a simple Excel table, and Figure 4.5 shows the
results of the TWR trend for different f values. Note that the highest value
64 is obtained risking 37% of the capital.
TWR
2.5
2.325110768
1.5
0.5
0
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
f% values
FIGURE 4.5 TWR trend for different f values.

37%
trade no. Result contracts equity
1 –500 29 85,500
2 1,300 25 118,000
3 700 34 141,800
4 –1,250 41 90,550
5 –1,000 26 64,550
6 2,500 19 112,050
7 5,350 33 288,600
8 –1,250 85 182,350
9 600 53 214,150
10 350 63 236,200
FIGURE 4.6 Results with f = 37%.
At this point, it’s important to field test an f of 37% to see how our 10
trades would have done. Figure 4.6 shows the results.
Everything in this section should be considered for purely educational
purposes, and I stress that it’s highly inadvisable for anyone to invest on the
markets using risk percentages close to optimal f. I must also say that the
above examples may be distorted by another parameter that hasn’t been
65
considered in this section: the margin of each single contract. The limit

of the number of contracts obviously depends on the margin required.
For example, if the margin was €2,500, the figures would probably hold
true anyway (we won’t run precise checks as this isn’t the purpose of this
chapter), but in cases of greater requirements it would have been impossible
to make many trades with the specified number of contracts.
■ 4.3 Secure f
The objections raised after analyzing the equity trends as the risk changes,
mainly concern the drawdown when taken as an indication of how much the
trader is suffering. Towards the end of the 1990s, Zamansky introduced the
concept of secure f as an alternative to optimal f.
In practice, this concept re-proposes the investor’s psychology as the cor-
nerstone on the basis of which market position choices are made; while
optimal f is based exclusively on recorded system data, with secure f the
trader takes a decision on the basis of the maximum historical drawdown.
So, of all the possible fractions, you choose the one that maximizes profits,
on the condition that the drawdown shown is below a certain percentage
chosen by the trader.
If we refer to Figure 4.2, a trader who didn’t want to find himself in
situations with a worse DD% than 25% would certainly choose an f of 15%
as his secure f. Another trader, more inclined to gambling, might be happy
with a DD% of 45% and as a consequence would choose a secure f of 25%.
On the other hand, if you didn’t want a DD% of over 4%, you’d choose a
percentage of around 2.5%.
Obviously, secure f will never be greater than optimal f, as the latter is the
value that maximizes system gains to the greatest extent.
If you’ve been following this chapter closely, you’ll see that there aren’t
considerable conceptual differences between the fixed fractional, optimal f,
and secure f methods, as they’re all based on the same principle of limiting
the risk of losing to a certain percentage of your capital. The optimal f and
secure f methods try to maximize profits in a way that’s compatible with the
system, in the first case, and with the investor’s psychology, in the second.
The fixed fractional method leaves you greater freedom, but its mathematical
construction is very similar to the other two.
66 With the fixed fractional method, I mentioned using the presumed
value of the loss to establish the number of contracts; in the examples
shown the theoretical system stop-loss was used. As mentioned in the

previous chapter, this is not always the best choice, as often the actual loss
is worse than what was expected due to market gaps, slippage caused by a
fast-market or technical problems with orders. Also referring to what Ralph
Vince used, Larry Williams proposed an application of the fixed fractional
method as a solution; this time, though, using the system’s maximum
historical loss. In his first works he even mentioned doubling said loss for
the calculations, although later he didn’t use this multiplication factor; but
it’s merely a quasi-philosophical question, as eventually the choice of the
percentage to risk is down to the trader, and using double the loss rather
than the actual loss could encourage the use of higher percentages, while it
would be more conservative to use just the actual loss.
In practice, using the fixed fractional method in a purely theoretical sense,
with a system that has a €1,000 stop-loss and a capital of €100,000, a trader
who wanted to risk 5% on each trade would start as follows:
5% of €100,000 = risk €5,000.
So, €1,000 loss on each contract for the first trade with five contracts
(in the worst-case scenario you’ll lose €5,000).
Williams, however, would have read the system report and found the
maximum historical loss, as he held that value to be more reliable than the
theoretical stop-loss value (this is a very valid reasoning, although you should
always bear in mind that it won’t protect you completely, as you might have
a worse trade than that in the future) and then he would have used that in
the calculation.
If, for example, the maximum historical loss was €1,200, without preju-
dice to the 5% risk of the available €100,000, the calculation would be
Contracts = 5,000∕1,200 = 4.167:
in other words, 4 rather than the 5 someone who took their decision based
on theory alone would have used.
As mentioned above, this approach is more prudent and it’s advisable to
use it even though it’s simply just a conceptual modification of the fixed frac-
tional method.
On the subject, Williams said: If you want to stay in the game, use a
conservative approach and don’t risk a lot; if you want to win trading cham-
pionships, use Kelly or optimal f.
According to some traders, who’ve survived various market ups and 67

downs, the concept of optimal f is useful to understand the absolute limit
beyond which you should never go if you want to avoid the risk of going
bankrupt, the point of no return called the cliff of death.
There are various facets to the fixed fractional method, the approach
described in this book is often called the percent risk model as you make trades
with a fixed risk percentage.
Another, much more banal, method used with futures, is to have one
contract for every €10,000 of available capital. The simplicity of this method
is obvious, and it’s also very simple to calculate the number of contracts. This
is exactly the same model as above, in which the risk percentage is:
( )
max loss
%f = − ∗ 100.
10,000
In fact, you can lose the maximum loss on each contract and, using this
with the capital of reference for that contract (€10,000 by definition for the
method), you obtain the percentage to risk (multiply by 100 to obtain the
percentage).
The banality of the method is as obvious as its riskiness; a maximum
(theoretical or real) system loss of €1,000 is the equivalent of a 10% risk
percentage, which can certainly be considered harsh.
■ 4.4 Fixed Ratio

Ryan Jones, in his work The Trading Game: Playing by the Numbers to Make
Millions, proposes an interesting alternative to the fixed fractional method
with his fixed ratio method.
According to Jones, the fixed fractional method reaches a point where
the increase in the number of contracts occurs too quickly and results in
drawdown levels that are unacceptable for any trader.
His work attempts to give each contract used the same weight, to maintain
its effect unchanged as equity increases.
He is quite damning of the fixed fractional method as, in his opinion, it is
too slow when increasing the number of contracts initially, and too fast once
capital has been accumulated, he says ironically after 10 years.
His reasoning goes more or less as follows:
Let’s suppose you have capital of €100,000, a %f of 10%, and a max. loss
68 of €5,000.
Said parameters would initially cause you to risk €10,000 (10% of
100,000) so you could use two contracts. In order to use three contracts,
you would need to reach €150,000; in fact 10% of 150,000 is €15,000,
which would let you use three contracts.
The next step would be €200,000, at which point you’d have four con-
tracts (10% of 200,000 is 20,000, which is four times the max. loss of
€5,000).
Note that the incremental single contract steps could be calculated as
follows:
step = max. loss∕%f
in fact: step = 5,000∕0.1 = 50,000.
With every 50,000 of profit you can increase your number of contracts by 1.
Ryan Jones argues (and does so quite logically) that’s it’s not the same
thing to increase your capital by €50,000 using just two contracts, as
at the beginning, as it is to do so once you’ve reached €1,000,000 and
are therefore using 20 contracts. In fact, in the first case, assuming you
make a profit of €2,500 per trade per contract, you’d need 10 trades
(2*2,500*10 = 50,000); while, in the second case, you’d just need one
trade (20*2,500 = 50,000).
Jones criticizes the slow, difficult start and then considers the increase
frenetic once you’ve built up a significant amount of capital. Again, according
to Jones, too sharp an increase at a certain point in the life of the system can
result in horrendous drawdowns.
Jones proposes a solution that increments the number of contracts: in
other words a ratio set between the gain per contract and the increase in the
number of contracts. A ratio of 1:5,000 means increasing by one contract
when every contract being used in the system has made €5,000 of profit.
The value of the increase per contract is commonly called delta.
Having set a value for delta, Jones increases the number of contracts
by one when the capital has increased by the value of delta for every con-
tract used.
An example is the best way to explain this concept:
Let’s take the previous case, with an initial capital of €100,000 and a max.
loss of €5,000. Jones doesn’t set a risk percentage, but rather a value for delta
that could be anything but which, for example, we’ll set as the same as the
max. loss; in other words €5,000.
As there is no initial rule, Jones starts trading with the system using just 69
one contract (note that in the %f example we started with two contracts due

to the characteristics of the system and the chosen risk profile).
In practice, with one contract it’s enough for the capital to reach €105,000
to start using two contracts; in fact every contract used (just one in this case)
will have made a profit of €5,000.
At this point you have two contracts and, according to the rule, to add
a third, each of the two contracts used must make a €5,000 profit, which
makes a total of €10,000.
So, from the €105,000 you reached you now have to reach €115,000 in
order to start using three contracts.
With three contracts you have to make a €15,000 profit (5,000 each con-
tract) in order to start using a fourth, so when you’ve past the €130,000
mark you can start using four up to €150,000 where, having made the nec-
essary €20,000 profit (four contracts times the €5,000 delta each) you can
add a fifth contract.
You’ll immediately note that, although you started with just one contract,
after the €150,000 mark you are already using five contracts, whereas with
Fixed
Ratio-delta
capital %f-10% 5,000
100,000 2 1
105,000 2 2
115,000 2 3
130,000 2 4
150,000 3 5
175,000 3 6
205,000 4 7
240,000 4 8
280,000 5 9
325,000 6 10
375,000 7 11
430,000 8 12
490,000 9 13
555,000 11 14
625,000 12 15
700,000 14 16
780,000 15 17
865,000 17 18
955,000 19 19
1,050,000 21 20
FIGURE 4.7 Different capital levels and numbers of contracts used on the basis of the
money management method chosen.
70
the fixed fractional method you started with two but after the €150,000
mark you only had three contracts.
Figure 4.7 shows the trend in the number of contracts using one method
compared to the other. Note that, in the initial phase the fixed ratio method
increases the number of contracts much faster, and is then passed by the fixed
fractional method.
It’s clear that a lower delta accelerates the increment in the number of
contracts even more, and this is therefore the variable the trader can use to
make the method more, or less, aggressive.
As a general guideline, Jones proposes using a delta equal to half the sys-
tem’s maximum historical drawdown, but leaves the trader free to choose
his own level of risk.
Note that using this method lets you trade in conditions in which other
methods have to limit exposure by ceasing trading. In fact, there isn’t a mini-
mum limit at which you should cease trading (unless you have no liquidity to
cover the margins set by your broker). In the above case, a 10% fixed frac-
tional method would be reduced to one contract if the capital dropped below
€100,000 and you would have had to cease trading if the capital dropped to
below €50,000. In fact, 50,000 * 0.1 = 5,000, which is equal to the max.
loss, and if less than this value (and it would be if the capital was less than
€50,000) you couldn’t trade with a max. risk of 10%.
With the fixed ratio method, on the other hand, there is only a level for
the first increment, but no stop level and theoretically you could continue
to use one contract for as long as margins allow. On the one hand, this is
certainly an advantage, but on the other, it might lead to some unpleasant
consequences if you can’t put a brake on your enthusiasm.
So, is there a formula for calculating the number of contracts for the fixed
ratio method, as there is for the fixed fractional method?
The answer is yes. I won’t go into detail about how they came up with the
formula, but I will show you how to use it:
If the capital < initial capital then contracts = 1
If the capital > initial capital then:
N.B.: In the following equations INT = ENT (entire part of)
71
contracts = ENT((1 + SQRT(1+8*(capital – initial capital)/delta)))/2)

or: √
⎛ (C − Ci) ⎞
⎜1 + 1 + 8 ∗ ⎟
⎜ delta ⎟
contracts = INT
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
Let’s take a look at a few examples to see how the equation works.
We’ll consider an initial capital of €50,000 and a delta of €2,500:
Ci = 50,000
delta = 2,500.
Starting with one contract, as mentioned above, once this contract made a
profit equal to delta we could start using the second contract. So in this case
we need to reach 52,500 to start using two contracts. Let’s see if that’s the
case:
√
⎛ (52, 500 − 50, 000) ⎞
⎜1 + 1 + 8 ∗ 2,500 ⎟
contracts = INT ⎜ ⎟
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
√
⎛ 2,500 ⎞
⎜ 1 + 1 + 8 ∗ 2,500 ⎟
= INT ⎜ ⎟
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
( √ )
1+ 1+8
= INT
2
( √ )
1+ 9 ( )
1+3
= INT = INT = INT(2) = 2.
2 2
From here on, each of the two contracts should make a €2,500 profit in
order to start using the third contract. This means we have to add another
€5,000 to our capital (2,500 for each contract used). So let’s check with the
72 equation:
√
⎛ (57, 500 − 50, 000) ⎞

⎜1 + 1 + 8 ∗ 2,500 ⎟
contracts = INT ⎜ ⎟
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
√
⎛ 7,500 ⎞
⎜ 1 + 1 + 8 ∗ 2,500 ⎟
= INT ⎜ ⎟
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
( √ ) ( √ )
1 + 1 + 24 1 + 25
= INT = INT
2 2
( )
1+5
= INT = INT(3) = 3.
2
The equation works, but the reader is free to run further checks to prac-
tice using the formula.
The average trade is the average trade of the system; on average, a contract
makes a profit on each trade equal to the average trade. If you divide the value
of delta by the average trade you get the average number of trades necessary
to increase the number of contracts used by 1. With the fixed ratio method,
in fact, you always add one contract when each of the contracts used has
gained delta.
Increasing the number of contracts at averagely constant intervals of
trades greatly increases the stability of the system, without producing
sudden fluctuations that could be difficult for the trader to withstand.
So which method is best? As usual, a lot depends on how much feeling
the trader has for the market. In certain cases, the choice of one method
over another is closely related to starting conditions. A very limited capital
in fact won’t let you effectively apply the fixed fractional method unless you
use very high risk percentages.
Let’s compare the fixed fractional and fixed ratio formulas used to calcu-
late the number of contracts:
( )
(C ∗ f )
ff contracts = INT
Maxloss
√ 73
⎛ (C − Ci ) ⎞
⎜1 + 1+8∗ ⎟

delta ⎟
FR contracts = INT ⎜ .
⎜ 2 ⎟
⎜ ⎟
⎝ ⎠
Note that, with the fixed fractional method, the number of contracts is
directly proportional to the capital (C in the equation), while with the fixed
ratio method, this number is calculated by taking the square root of the
capital.
This means that the increase in the number of contracts is more marked in
the first case, except at low capital values. As mentioned above, the number
of contracts increases faster in the initial phases of the system using the fixed
ratio method, to later be passed by the fixed fractional method.
This can be interpreted in two diametrically opposed ways.
The first school of thought condemns the behaviour of the fixed ratio
method, as it risks more when you have less, and risks less when you have
accumulated significant capital.
On the other hand, those who defend the fixed ratio method say this
behaviour is more than simply the only one you can use with a reduced cap-
ital; it also produces a much more stable equity line once the capital has
reached higher values.
Let’s take another look at our euro/dollar system, with various scenarios
resulting from the application of one method or the other, and the limits of
first one, then the other.
Figure 4.8 shows the equity produced by applying the fixed fractional
method with a 5% risk percentage and an initial capital of €50,000 (for prac-
tical reasons from here on we’ll stick to euros instead of dollars, even though
the currency is really USD; this doesn’t change the results in conceptual
terms).
Many might think a 5% risk percentage isn’t very aggressive, but with an
available capital of €50,000, this means risking €2,500, which can be con-
sidered acceptable. In actual fact, it’s not a single loss that’s the psychological
problem for a trader; it’s a sequence of negative trades. The equity shown has
the same initial flat period, then increases rapidly, as we saw in the examples
in which we applied the Kelly formula.
The final result is notable, with a capital of €654,917.5, but the maximum
drawdown of €350,000 from 639,920 to 275,317.5 is also considerable!
74
Figure 4.9 shows how a trader would have done, again starting with
€50,000, if he risked 10% of his capital on each trade.

The result is impressive to say the least: the upswing reaches incredible
highs, but the subsequent crash from €3,173,825 to just €476,732.5 is also
incredible (and, in my opinion, intolerable).
f% = 5%
800,000.00
639,920.00 654,917.50
600,000.00
400,000.00
275,317.50
200,000.00
0.00
1 41 81 121 161 201 241 281 321 361 401 441 481 521 561 601 641 681 721 761 801 841 881 921 961
FIGURE 4.8 Applying a 5% fixed risk percentage.

f% = 10%
3,400,000.00
3,173,825.00
3,200,000.00
3,000,000.00
2,800,000.00
2,600,000.00
2,400,000.00
2,200,000.00
2,000,000.00 1,967,672.50
1,800,000.00
1,600,000.00
1,400,000.00
1,200,000.00
1,000,000.00
800,000.00
600,000.00
400,000.00
495,567.50
200,000.00
0.00
1 42 83 124 165 206 247 288 329 370 411 452 493 534 575 616 657 698 739 780 821 862 903 944
FIGURE 4.9 10% risk on each trade.
It looks like we’ve found another goose that lays golden eggs. What would
happen if we used 11% instead of 10%, just 1% more?
The results are shown in Figure 4.10. 75

f% = 11%
200,000.00
11,235.00
0.00
1 41 81 121 161 201 241 281 321 361 401 441 481 521 561 601 641 681 721 761 801 841 881 921 961
FIGURE 4.10 A risk percentage that’s just one point higher blocks the system.
The same thing we saw before (with the Kelly formula) occurs. The resid-
ual capital can only be used in theory, but not in practice.
With the fixed fractional method we set a hypothetical max. loss of
€1,250, and calculated the number of contracts on this basis.
At €11,235, the calculation is:
( )
(11, 235 ∗ 0.11) 1, 235.85
contracts = ENT = ENT = 0.
1,250 1,250
This may seem slightly puzzling to some who may make the following
consideration: if with 11% I can only risk €1,235.85 and then have to cease
trading, what could I do with 10%, which is even less? The answer is very
simple, risking 10% of the capital from the start you would have reached
the same point in the system with more residual funds. In particular, the
trade that, with 11%, reduced the capital to €11.235 was trade number
129. For that same trade, starting with a 10% risk, the available capital was
still 13,547.5, and the result would be:
( )
(13, 547.5 ∗ 0.10) 1,354.75
contracts = ENT = ENT = 1.
1,250 1,250
76
Just a small change makes a big difference in the end.
Note that for this system optimal f would be 23.4%, but you obviously
wouldn’t be able to continue using this system with such a high percentage;
calculated at a purely theoretical level, the only difference being the remain-
ing capital when you threw in the towel, even less than with f = 11%, as
shown in Figure 4.11.
So, just as percentages that are too high end in a sort of bankruptcy, per-
centages that are too low could also prevent you from continuing, or even
starting, to use the system.
With an available capital of €50,000 and a max. possible loss of €1,250,
obviously the fraction risked must be greater than €1,250 in order to start
trading. €1,250 of 50,000 is 2.5%, so this is the minimum possible per-
centage you could use for this system, at least to start trading. In actual
fact, already after the second trade, which reduces the equity to less than
€50,000, you will have to cease trading.
f% = 23.4% - optimal f
200,000.00
4,752.50
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.11 An attempt using optimal f.
With a little trial and error, we can establish that the minimum percent-
age necessary to continue trading to the end is 3.75%, and this risk would
produce the equity in Figure 4.12. 77

f % = 3.75%
400,000.00
297,632.50
293,025.00
200,000.00
158,207.50
0.00
1 41 81 121 161 201 241 281 321 361 401 441 481 521 561 601 641 681 721 761 801 841 881 921 961
FIGURE 4.12 Minimum risk percentage to keep trading.

Obviously, this percentage depends on the initial capital in the same way
as the initial capital depended on the result of the approach when using the
Kelly formula.
Note that, even with the minimum possible percentage, the drawdown
is still significant. We should remember, however, that this system wasn’t
considered to be good in the original report (Figure 3.2) and we’re only
using it in these examples to show the effect of applying money management
strategies.
Now let’s take a look at the results we would have obtained using the fixed
ratio method.
In this case, it’s the choice of delta that acts as an amplifier for the number
of contracts. The smaller delta is, the faster more contracts can be used as
equity increases.
For the first test, let’s try a delta that’s approximately 50% of the max.
system historical drawdown. As shown in Figure 3.2, this drawdown is
approximately €22,000, so we’ll start with a delta of €11,000. This means
we’ll add a contract every time each of the contracts used has made €11,000
of profit for the system.
Figure 4.13 shows the results obtained.
The results look quite promising, and there are some similarities to the
78 fixed fractional method line using a minimum risk.
Considering a delta of €11,000 to be rather high, let’s take a look at the

results with a delta of €5,000 and €2,500, the first acceptable the second
very aggressive (these figures weren’t chosen at random and are respectively
four times and double the maximum theoretical loss).
As you can see in Figure 4.15, the results are better than those obtained
using the fixed fractional method with f = 5% (Figure 4.8), but the draw-
down is about the same. Actually, the trend of the drawdown with this sys-
tem doesn’t look good for the fixed ratio method either and, if there’s an
advantage to be found using this method rather than the fixed fractional
method, it lies in the lack of lower risk limits. While in fact, using the fixed
fractional method you can’t go below certain levels, with the fixed ratio
method there are no limits below which the system can’t be used.
In the same way, using the fixed ratio method you can push things beyond
the higher risk levels, as there’s the theoretical possibility of trading with a
single contract for as long as margins allow.
delta = 11,000
400,000.00
344,430.00
312,822.50
300,000.00
200,000.00
186,885.00
100,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.13 Fixed ratio with delta = €11,000.
79
delta = 5,000

600,000.00
500,000.00 508,892.50
466,812.50
400,000.00
300,000.00
236,505.00
200,000.00
100,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.14 Delta four times the maximum theoretical loss.

delta = 2,500
1,000,000.00
900,000.00
869,507.50
800,000.00
790,547.50
700,000.00
600,000.00
500,000.00
400,000.00
365,035.00
300,000.00
200,000.00
100,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 4.15 Delta double the maximum theoretical loss.
80
delta = 1,250
1,500,000.00
1,400,000.00
1,300,000.00 1,303,430.00
1,196,145.00
1,200,000.00
1,100,000.00
1,000,000.00
900,000.00
800,000.00
700,000.00
600,000.00
500,000.00
463,995.00
400,000.00
300,000.00
200,000.00
100,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 4.16 Delta equal to the maximum theoretical loss.

delta = 625
300,000.00
224,637.50
200,000.00
100,000.00
48,910.00
39,672.50
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.17 A delta equal to half the maximum theoretical loss is too aggressive.
An aggressive strategy that pays off (in this specific case) is the one that
uses a delta equal to the maximum theoretical loss – in other words, €1,250.
The results are shown in Figure 4.16.
You can see some interesting results, with the trend of the drawdown on a 81

par with previous cases, rather than the unsustainable behaviour (compared
to the other risk percentages) seen in Figure 4.9.
Finally, here’s a limit case, in which the risk becomes excessive and the
system fails, as shown in Figure 4.17.
Note that, in this case, the equity never really takes off – except at
the system’s best moment, when the capital is increased fourfold (max.
224,637.50). But then, the excessive risk reduces the equity to below the
initial levels, where it stays to the end.
Note that, in this case, a particularly benevolent future phase of the mar-
ket wasn’t excluded, and this gives the equity a considerable boost, while
with the fixed fractional method the system is brought to a halt for good.
■ 4.5 Percent Volatility Model

This method is more difficult to apply in terms of constructing the equation
used to calculate the number of contracts.
The idea behind it, as clearly described by Van K. Tharp in Trade Your
Way to Financial Freedom, is based on market fluctuations and the maximum
fluctuation you wish to withstand with your capital.
To measure the fluctuations of the instrument you intend to trade with,
a measure of the volatility of the same is used; the daily range from max. to
min. is measured, and an average is calculated after a certain number of days.
In general, a few days are considered for systems that don’t have a long-term
open position, and more days are considered for more long-term systems.
Let’s say it’s reasonable to use an average of four to five days for quite fast
systems (positions open from two days to one week) and 20 days for systems
that follow a trend for longer (positions open from 10 days to one month).
The average described is the average of the ranges in the above period. In
actual fact, it’s more convenient to consider the average true range (ATR)
in order to include a possible gap in the opening prices.
In order to understand what the ATR is, we should hypothetically con-
sider a daily futures’ bar on the SPMIB index (previously called FIB), the
range of which goes from 32,100 to 32,450. The range of this bar is 350
points, or €1,650, as the value of each point is €5. If said bar showed a gap
up 50 points on the close of the previous day and the close was 32,050, the
82 true range would allow for this gap and would measure 350 + 50 = 400
points, or €2,000. What’s more, as the position will be open for several
days, it’s only right to consider cases in which the bars were all above or
below the close of the previous day.
Figures 4.18 and 4.19 show the two cases in which the bar shows gaps
above or below the close.
Now, going back to our model, let’s calculate the average of the last five
true ranges and take said value as the period volatility. For example, the value
may be 300 points, or €1,500.
Supposing we have an available capital of €100,000 and want to set a maxi-
mum fluctuation of 2%. The allowed fluctuation would therefore be €2,000.
Considering the average volatility of the future calculated as €1,500, the
number of contracts is calculated as follows:
( )
2,000
contracts = ENT = 1.
1,500
If Vp is the value of the single instrument point (in our example, Vp
= 5) and Vol% is the allowable fluctuation percentage (in our example,
RANGE
TRUE RANGE
RANGE = TRUE RANGE
FIGURE 4.18 True range in case of low above previous day’s close.
83

RANGE = TRUE RANGE RANGE TRUE RANGE
FIGURE 4.19 True range in case of high below previous day’s close.
Vol% = 2%), the number of contracts would be:

( )
Vol% ∗ capital
contracts = ENT .
ATR(period) ∗ Vp
It’s immediately obvious that the construction of this method is more
sophisticated as you need to know the true ranges of a certain number of days
before the entry point. This value can be calculated when making the trade,
but this makes back-testing the system more difficult unless you export a
larger amount of data to Excel, or enter the equation for calculating the
number of contracts directly in the EasyLanguage code.
The method is of a certain interest as it’s not linked, in its own particular
way, to the worst loss suffered, although it comes to a halt in conditions con-
sidered excessive. Note that, unlike what occurred with the fixed fractional
system, which came to a definitive halt once you arrived at an insufficient
capital for the maximum allowed loss, this method can be put on standby for
periods of time and then start following the signals again in the right volatil-
ity conditions. A decrease in market volatility in fact increases the number
of contracts calculated by the equation (the volatility is the denominator of
the equation, which is a fraction; therefore, the smaller it is, the higher the
result) and, at low volatility levels, there may be conditions for following a
trade even with very little capital.
Note that this is perfectly logical, if volatility is low you risk less and with
less risk you can also trade with less capital.
By way of example, we calculated the results of the euro/dollar system
also applying the % volatility method.
Here are the results.
Let’s start with 5%, which means you don’t want your capital to be
affected by volatility over 5%.
Figure 4.20 shows the resulting equity.
84
%vol = 5%
1,000,000.00
900,000.00
800,000.00
761,460.00
700,000.00
670,257.50
600,000.00
500,000.00
400,000.00
300,000.00
243,997.50
200,000.00
100,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 4.20 Results attempting to limit the volatility of your capital to less
than 5%.
%vol = 10%
2,300,000.00
2,181,435.00
2,200,000.00
2,100,000.00
2,000,000.00
1,900,000.00
1,800,000.00
1,700,000.00
1,600,000.00
1,500,000.00
1,400,000.00
1,300,000.00 1,311,575.00
1,200,000.00
1,100,000.00
1,000,000.00
900,000.00
800,000.00
700,000.00
600,000.00
500,000.00
400,000.00
300,000.00
200,000.00
216,257.50
100,000.00
0.00
4
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 4.21 Results attempting to limit the volatility of your capital to less
than 10%.
Figure 4.21, on the other hand, shows the results of limiting volatility to
10%.
85
As this is conceptually similar to the fixed fractional method, we should

of course compare these results with that method and the risk percentages
used for the same.
A 5% f% produced the results in Figure 4.8, a final equity of €654,917.5
and a drawdown from €639,920 to €275,317.5. Note that with a %Vol of
5% the drawdown is very similar to that in Figure 4.8, while the final result
is approximately €130,000 higher. Therefore, for the same sufferance the
%Vol method is best in this case.
At higher percentages such as 10% the %f gives better results, with the
equity reaching almost 2 million euros (Figure 4.9), but the drawdown is
also incredible, from almost 3 million euros to just under half a million.
Using %Vol the final result is €1,311,575 but with a lower drawdown, if
you can call a drop of almost 2 million euros low.
Remember how more aggressive traders were already brought to a halt
by the fixed fractional method risking 11% of their capital on each trade
(Figure 4.10)? Using %Vol, on the other hand, in this case, you can go much
further, as the results in Figure 4.22 show, although I doubt anyone would
want to take this as an example for their own trading.
%vol = 15%
3,300,000.00 3,142,717.50
3,200,000.00
3,100,000.00
3,000,000.00
2,900,000.00
2,800,000.00
2,700,000.00
2,600,000.00
2,500,000.00
2,400,000.00
2,300,000.00
2,200,000.00
2,100,000.00
2,000,000.00
1,900,000.00
1,800,000.00
1,700,000.00
1,600,000.00
1,500,000.00
1,400,000.00
1,300,000.00
1,200,000.00
1,100,000.00
1,000,000.00
900,000.00
800,000.00
700,000.00
600,000.00
500,000.00 468,662.50
400,000.00
300,000.00
200,000.00
100,000.00 67,022.50
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 4.22 %Vol allows the use of more aggressive, although not exactly
recommendable, percentages.
It’s only with even higher percentages that the system goes bankrupt, and
86 in this case, bankrupt is exactly the right word as the residual capital with a
%Vol of 20% would be that shown in Figure 4.23.
With the fixed fractional method, we saw that, depending on the initial
capital, not all the risk percentages let you trade. Too low a risk often left
insufficient capital, and Figure 4.12 showed how 3.75% was the minimum
possible percentage you could use without being brought to a halt.
With %Vol, however, the limits are lower simply because there will always
be periods in which the market is less volatile, when you can trade also with
less capital. Figure 4.24 shows the results, with a percentage that’s near the
minimum that could be used.
In order to better comprehend how some trades are missed, let’s take a
look at the table of trades in Figure 4.25, using an allowed volatility of 2%.
After the first trade, possible because the market volatility measured with
a five-day ATR was €924.99, if you are willing to risk 2% of the initial
€50,000, in other words €1,000, the number of contracts is:
( 1,000 )
Contracts = ENT =1
942.99
%vol = 20%
100,000.00
2,112.50
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.23 Too high a risk quickly reduces the capital to a minimum.
87
%vol = 2%

300,000.00
200,000.00
187,075.00
164,080.00
117,767.50
100,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 4.24 Using %Vol you can even proceed with just a 2% risk.
trade volatility %vol contracts %vol cumulative 2% of equity
64,250 92,499 1 5,064,250 101,285
–172,000 105,500 0 5,064,250 101,285
–3,250 109,000 0 5,064,250 101,285
9,250 111,250 0 5,064,250 101,285
–85,750 110,750 0 5,064,250 101,285
48,000 109,000 0 5,064,250 101,285
–150,750 103,000 0 5,064,250 101,285
–127,000 105,000 0 5,064,250 101,285
–215,750 130,000 0 5,064,250 101,285
159,250 138,500 0 5,064,250 101,285
8,000 104,750 0 5,064,250 101,285
–127,000 93,250 1 4,937,250 98,745
–2,000 118,750 0 4,937,250 98,745
FIGURE 4.25 First trades using an allowed volatility of 2%.
from this point on, the volatility is always above the level of the maximum
range which, recalculated for the level of equity after the first trade of
€50,642.5, is €1,012.85. You can only enter the market with a contract
when the volatility drops to 932.498. Unfortunately, this is a losing trade
and the volatility increases once again, immediately bringing the system to
a halt once more.
After running a few tests, there were six cases in which, three pairs
88 showed a similar performance in terms of percentage returns. Of these,
the maximum percentage drawdown was measured to see which of them
caused the most difficulty to the trader who chose them.

Figure 4.26 shows a table with a summary of these results.
As expected, the fixed ratio method is the one which, for the same
returns, has a lower drawdown percentage. It’s also interesting to note how
apparently the % volatility method produces better results with less risk
exposure.
fixed fractional final equity equity % max DD%

4% 315,070 530% 46.50%
7% 1,256,233 2412% 71.30%
FIXED RATIO
12,500 316,003 532% 38.70%
1,260 1,254,973 2410% 62.00%
% volatility
3.15% 319,855 540% 44.10%
5.60% 1,262,135 2424% 68.90%
FIGURE 4.26 Similar returns using different approaches.

%f trend
2,500,000.00
2,000,000.00
1,500,000.00
1,000,000.00
500,000.00
0.00
1. %
2. %
4. %
5. %
6. %
7. %
9. %
10 0%
11 0%
13 0%
14 0%
15 0%
17 0%
18 0%
19 0%
20 0%
22 0%
23 0%
24 0%
0%
10
40
70
00
30
60
90
2
.5
.8
.1
.4
.7
.0
.3
.6
.9
.2
.5
.8
0.
FIGURE 4.27 Final equity trend as the risk percentage changes.
Fixed Ratio trend

2,000,000.00
1,800,000.00
1,600,000.00
1,400,000.00
1,200,000.00
1,000,000.00 89
800,000.00

600,000.00
400,000.00
200,000.00
0.00
1.2 0
2.3 0
3.4 0
00
5.6 0
6.7 0
00
8.9 0
10 00
11 00
12 0
13 00
14 00
15 00
16 00
17 0
18 00
19 0
0
10
0
0
0
0
.10
.60
.80
.90
4.5
7.8
.0
.2
.3
.4
.5
.7
FIGURE 4.28 Trend of the equity as delta changes.
For the system being used, it may be a good idea to create charts of the
final equity results, each time, for the percent f, fixed ratio, and percent
volatility methods.
Figure 4.27 shows the end result for the fixed fractional method as f
increases. Figure 4.28 shows the trend as delta increases using the fixed ratio
method. Figure 4.29 shows the equity trend as the maximum percentage
changes using the percent volatility method.
Note the evident differences in the construction of the lines.
%vol trend
2,500,000.00
2,000,000.00
1,500,000.00
1,000,000.00
500,000.00
0.00
0%
0%
0%
0%
0%
10 %
12 %
13 %
15 %
17 %
18 %
20 %
22 %
23 %
25 %
27 %
29 %
%
0
.30
.00
.70
.40
.10
.80
.50
.20
.90
.60
.30
.00
0.1
1.8
3.5
5.2
6.9
8.6
FIGURE 4.29 Trend of the equity as the maximum allowed percent volatility changes.
The line in Figure 4.27 starts flat, and considering the low risk adopted,
it’s practically impossible to start using the system. Then there is a notable
increment as the risk increases until quite an irregular high area. Note that
around 10%, there are percentages that produce the highest profits and per-
90
centages that lead to bankruptcy. If we take another look at Figure 4.5, the
line is very clean with a very regular curvature. Figure 4.27, on the other
hand, looks nothing like it, despite being the same method. The difference,
once again, is due to the application of a theoretical method in the real world.
Figure 4.5 was created without considering market limits, while Figure 4.27
considers everything, from the necessary rounding off of the calculations, to
the use of a whole number of contracts, to the limits imposed by margins. In
Figure 4.5, the optimal value for maximising profits is 23.4%. In actual fact,
you can’t use more than 10% because you can’t use a fraction of one contract
to continue following the signals of the system when the capital is reduced
after the first losses (the possibility of using mini futures or something similar
isn’t considered as it lies outside this context).
The line shown in Figure 4.28, however, shows what happens to the
final result using the fixed ratio method. There’s an almost vertical upswing
immediately (as mentioned above, the lower the delta the more aggressive
the system), just over the values that made the system go haywire (see the
test with a delta of 625). Then there’s a slow but gradual decline with a few
ups and downs caused by rounding off to whole numbers of contracts.
Those who wish to avoid sudden fluctuations in returns can use this
method, taking care in the initial period or in any period in which the
equity drops below the initial capital, with a system that lets you continue
in any case with just one contract, as long as the margins allow.
Note that the highest profit that can be obtained using this method is less
than the other two cases, although not a lot less.
Figure 4.29 shows the line produced by the limit imposed on the fluctu-
ations in the equity on the basis of market volatility, taken as the average of
the last five true ranges.
This line is also quite erratic around the peak percentages. Unlike the
line produced by the fixed fractional method, however, there is a much less
marked drop, which produces a line similar to the classic ‘Gaussian’ line.
In practical terms, this could mean a brighter future for the trader who,
willingly or by mistake, went too far with the percentages (note that, using
the fixed fractional method, beyond a certain point there’s nothing more
that can be done). Obviously, with both the fixed fractional method and the
percent volatility method, there are percentages above which there are only
losses. The charts in fact don’t go as far as 100%, and show a lot less than half
that. Also, the fixed ratio method has its limits at much lower delta values,
but in a much narrower band than the other methods.
91

■ 4.6 Levels for Changing the Number
of Contracts
The formulas used in the book to calculate a suitable number of contracts for
the chosen risk profile can obviously be used for trading; but, in principle,
one could also calculate the various threshold levels at which to increase or
decrease the number of contracts being used.
Let’s take the fixed fractional method, for example, with a maximum the-
oretical loss of €1,250 and a 5% risk, it’s immediately evident that with this
method you’ll use one contract for every €25,000. The equation is:
( )
(C ∗ f )
ff contracts = INT .
Maxloss
If we use Contracts = 1 we obtain the inverse formula:
Maxloss 1,250
C= = = 25,000.
f 0.05
In practice, one contract is added, every €25,000.
TABLE 4.3
Capital Contracts
< 25,000 System stopped

>=25,000 and < 50,000 1 contract
>=50,000 and < 75,000 2 contracts
>=75,000 and < 100,000 3 contracts
>=100,000 and < 125,000 4 contracts
... ...
>=500,000 and < 525,000 20 contracts
TABLE 4.4
Capital Contracts
< 55,000 1 contract

>=55,000 and < 65,000 2 contracts
>=65,000 and < 80,000 3 contracts
>=80,000 and < 100,000 4 contracts
>=100,000 and < 125,000 5 contracts
92
It’s simple, one can trade on the basis of Table 4.3.

In the same way you can calculate the fixed ratio levels for a known
delta value. With a delta of €5,000 and one contract to start trading with
€50,000, by definition, you’d add the second contract at €55,000, the third
at €65,000, and so on. So, as shown in Table 4.4.
In this case, the following threshold is again calculated by adding the value
of delta for the number of existing contracts.
It can be a good idea to create a table like this so you can quickly see the
number of contracts to use for the trade you’re opening.
■ 4.7 Conclusions
We’ve considered various approaches that can be used to manage the num-
ber of contracts, some of which are based on the maximum theoretical loss
or real past loss of the system, and on this basis you can try to limit future
losses to a percentage of your available capital. These methods are based on
a risk percentage; in other words how much of your capital you’re willing to
lose in the trade you’re about to place. Another system attempts to increase
exposure more gradually once the initial capital has increased, and by con-
sidering all existing contracts to be equally important, in practice invests
the profits made on the market equally. Finally, the last approach considered
assesses recent market behaviour, and assuming the same levels of volatility
may be maintained in the short term, aims to limit fluctuations in the equity,
limiting the rise or fall of percentage values.
Note that in all the proposed methods, an aggressive approach only pays
off up to a certain point. When there is a considerable increase in the per-
centage drawdown values as risk increases, you reach a failure point at which
you lose most of your capital after the initial exponential gains.
■ References
Jones, Ryan. The Trading Game: Playing by the Numbers to Make Millions. New
York: John Wiley & Sons, 1999.
Tharp, Van K. Trade Your Way to Financial Freedom. New York: McGraw-Hill,
2007. 93
Vince, Ralph. Portfolio Management Formulas. Mathematical Trading Methods for

the Futures, Options, and Stock Markets. New York: John Wiley & Sons,
1990.
CHAPTER 5
Refining the
Techniques
■ 5.1 The Importance of the Trader’s
Temperament
94 Once you’ve familiarised yourself with the techniques described in the
previous chapter, with due consideration you can go beyond the basics and
study the variations that are best suited to your character.
As we’ve seen, there are different schools of thought on the exposure to
risk, depending on the capital you’ve accumulated. Ryan Jones with his fixed
ratio method prefers a more noticeable risk at first, then a subsequent reduc-
tion when you’ve accumulated significant capital. Everything is relative; a
trader who starts with €10,000 could tolerate a €500 loss, which would be
5% of his capital, but he’ll probably consider a €5,000 loss hair-raising. If
the same trader used the fixed fractional method, and everything went as
planned, at a certain point he might have increased his capital to €100,000.
At that point risking the same 5% would mean risking the same €5,000, but
now it would probably worry him less. What we’re doing is analyzing things
from the point of view of the person’s temperament. If the trader had made
that €100,000 over a period of five years, he’d have gradually got used to
larger and larger cash flows, and would be naturally ready to face a potential
loss of €5,000. If the increment was unexpected, however, and the lucky
trader found himself with €100,000 in just six months, I sincerely doubt
that, no matter how euphoric he was, he’d be psychologically prepared for
the ‘new’ situation and €5,000 would still seem like a lot.
My intention, with this example, is to get you to reflect on the different
positions someone who chooses one or the other route will be in. He who
sets a constant %f won’t worry about how much available capital he has and
how long it takes the situation to change, continuing unruffled to mechani-
cally apply the chosen method. He who thinks along the same lines as Jones,
on the other hand, will consider it wiser to reduce the risks at a certain point
to avoid heavy losses.
The choice between the fractional f or fixed ratio method (leaving per-
cent volatility aside for a moment) isn’t the only way to match the trader’s
temperament, though.
■ 5.2 Reduced f
If, for example, we decided to trade with a %f of 10% until our capital dou-
bled, then reduced it to 7.5% until it was three times the initial capital, and
95
then set the risk percentage to 5% until it was four times as much, to 2.5%
REFINING THE TECHNIQUES

up to eight times the initial capital and then 1.25% for every higher value,
we’d be trading more aggressively at the start, then reducing our exposure
more or less gradually (the steps were chosen merely by way of example).
Table 5.1 shows a summary of the chosen steps.
To check the validity of this approach, we’ll compare it with the equity
that produces more or less the same result, adopting the fixed ratio money
management method.
TABLE 5.1
equity %f
<= 2 x initial capital 10%

>2 and <= 3 x initial capital 7.5%
>3 and <= 4 x initial capital 5%
>4 <= 8 x initial capital 2.5%
> 8 x initial capital 1.25%
Fixed Ratio Vs Reduced f
600,000.00
477,912.50
459,982.50
400,000.00
Fixed Ratio delta 6,000

reduced f
200,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
1
FIGURE 5.1 The final results are almost the same but with different trends.
Figure 5.1 shows a direct comparison with the line produced using a delta
of €6,000.
The bold line shows the line produced by the reduced percentage
method, while the thin line shows the line produced using the classic fixed
ratio method. The idea of the approach to risk is the same in both cases.
96 In fact, both gradually reduce how aggressive the method is as the capital
increases. Note, though, how the reduced f method (as I’ve arbitrarily
named it) has a ‘softer’ trend and the drawdown suffered around halfway is
that of the fixed ratio line.
Note also that in the initial period, the reduced f line dropped below the
fixed ratio line. In fact, this is the period in which it adopted an aggressive
approach with a 10% risk, which accentuates negative periods.
Many variables can be used when choosing steps to build a risk model
that’s well-suited to any trader.
Let’s look at another example, with more aggressive percentages, in
Table 5.2.
As well as the final higher percentage, ‘stronger’ percentages are used
throughout this example.
This choice produces a final capital of €623,827.5. To obtain a comparable
result using the fixed ratio method, you’d have to adopt a delta of €3,500.
Figure 5.2 shows a comparison of the lines.
TABLE 5.2
equity %f
<= 2 x initial capital 10%

>2 and <= 3 x initial capital 9%
>3 and <= 4 x initial capital 7.5%
>4 <= 6 x initial capital 5%
> 6 x initial capital 2.5%
Fixed Ratio Vs reduced f

800,000.00
633,137.50
600,000.00
623,817.50
Fixed Ratio delta 3,500

400,000.00
reduced f
200,000.00
97

0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.2 Another comparison confirms the fluctuations in the equity line with the
reduced f method are less harsh.
Also in this case we can see that, while both lines show a more
pronounced dip in the drawdown period, the reduced f method loses ‘just’
around €200,000 compared to 300,000 with the fixed ratio method.
■ 5.3 Aggressive Ratio

Obviously, we could apply the same principle, inverted, to the fixed ratio
method to make it more aggressive after accumulating sufficient funds.
But what we’re doing is trying to limit the drawdown and make the capital
growth line as tolerable as possible; applying a strategy that decreases delta
as the capital increases would be a move in the opposite direction.
Merely by way of example, let’s see how a fixed ratio strategy would
behave. In this case, we’ll call aggressive ratio that reduces delta by 10%
every time a profit equal to the initial capital is made. In any case, we’ll limit
the minimum delta used to a quarter of the initial delta. The values chosen
are absolutely arbitrary and the purpose of the same is merely to show the
trend of the resulting line.
Table 5.3 shows a summary of the delta used, starting from 5,000, with
an initial capital of €50,000.
Figure 5.3 shows the equity compared with an equity that produces a
similar final result obtained applying the fixed fractional method (with a risk
of 5%).
You’ll immediately see that an aggressive ratio strategy doesn’t produce
any advantages as the equity line looks just like the %f one. The drawdown in
the aggressive ratio version is slightly worse than that of the corresponding
fixed risk line, so there’s no point in looking for special algorithms to vary
delta to make the method more aggressive when equity is high.
98
TABLE 5.3
Equity Delta
From 50,000 to 100,000 5,000

From 100,000 to 150,000 5,000 * 0.9 = 4,500
From 150,000 to 200,000 5,000 * 0.8 = 4,000
From 200,000 to 250,000 3,500
From 250,000 to 300,000 3,000
From 300,000 to 350,000 2,500
From 350,000 to 400,000 2,000
From 400,000 to 450,000 1,500
From 450,000 to 500,000 1,250
From 500,000 up 1,250
Aggressive Ratio Vs Fractional f
800,000.00
658,470.00
654,917.50
639,920.00 622,087.50
600,000.00
Fractional f
aggressive Ratio
400,000.00
275,317.50
200,000.00
220,842.50
50,642.50
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.3 A comparison between a classic %f line and a Fixed Ratio line with a
decreasing delta (aggressive ratio).
■ 5.4 Asymmetric Ratio

In his paper, Jones proposes a variation of his method called asymmetric ratio. 99

In this version he studies the possibility of changing the levels at which a
contract is added or subtracted as equity increases or decreases. His first
consideration is to decrease the number of contracts more quickly when
equity decreases to limit the possibility of a drawdown.
This is certainly a good idea, in order to limit damage while still obtaining
good results, but it is complicated to implement. Every method someone
comes up with needs to be tested, and there’s nothing to say that what works
well on one system will work well on others.
In order to better understand Jones’s idea let’s take a delta of 5,000 and
an initial capital of €50,000. At €55,000, you’d start using two contracts,
and three at 65,000. With a symmetrical method, below 65,000 you’d go
back to two contracts and below 55,000 to a single contract. Adopting an
asymmetrical method on the other hand you could decide to go back to one
contract halfway, or go from two to one at 60,000 instead of 55,000 (60,000
is halfway between 55,000 and 65,000). A test run on our system shows this
doesn’t produce real benefits, though. The drawdowns, in fact, remain more
or less the same, and final profits are slightly lower.
Asymmetric Ratio Vs Fixed Ratio
600,000.00
508,892.50
480,522.50
400,000.00 Asymmetric Ratio

Fixed Ratio
200,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.4 A comparison between the fixed ratio method and a version taking things
slow to keep everything under control.
Figure 5.4 shows the comparison between the two lines. It’s easy to see
that these complications in terms of calculations didn’t produce the benefits
100 we’d hoped for.
■ 5.5 Timid Bold Equity

The above test was run in an attempt to try and ‘calm’ the roller-coaster
ride of an equity line in relation to the increase in the number of contracts.
Obviously, there are also those whose reasoning is exactly the opposite,
who are happy to safeguard the initial capital and risk all the profits they’ve
made. In a certain sense, the fixed ratio method behaves intrinsically in this
way, increasing the number of contracts only as profits increase. In fact, the
formula considers the difference between the level of the equity and the
initial capital, therefore excluding the latter when calculating the number of
contracts.
The fixed ratio method, on the other hand, has proven to be a rather
‘conservative’ method as your available capital increases, and might not
be the ideal solution for someone who wanted to use a more aggressive
approach on the market, while still treating your initial capital with kid
gloves.
On this subject, K. Van Tharp proposes the idea of splitting your capi-
tal into two parts, the first to be traded at a low risk and the second at a
high risk (which he calls timid and bold). His idea is to calculate the num-
ber of contracts to use with the fixed fractional method, but applying two
different percentages to the two parts of the capital, in particular a low-risk
percentage to the part to keep and a high-risk/returns percentage to the
remaining part.
In particular, he considers it expedient to apply the optimal f percentage
to the high-risk part.
Finally, as far as the division is concerned, he suggests considering the
initial capital low risk and the profits made on the market high risk. This can
be done using any thresholds you want. In particular, once the capital has
doubled, it would be advisable to start again from scratch, keeping double
the initial capital that was kept previously and applying the risk percentage
to what’s left over.
Starting with €50,000, for example, you’d trade with a percentage of 5%
applied to €50,000, the result of which is 2,000. To this you’d add the value
you get from applying the optimal percentage to the profits. Remember,
optimal f for this system was 23.4%. Therefore, if you reached a capital of
€60,000 you’d calculate the number of contracts as follows:
( ) 101
((50, 000 ∗ 5%) + (60, 000 − 50, 000) ∗ 23.4%)

contracts = ENT
1,250
= ENT(3.872) = 3.
On the other hand, by applying 5%, as we would traditionally, to the
entire €60,000, we would have obtained:
( )
(60, 000 ∗ 5%)
contracts = ENT = ENT(2.4) = 2.
1,250
After €100,000, you’d apply 5% to 100,000 and 23.4% to anything over
that, up to 150,000 where the capital up to 150,000 would be calculated at
5%, and so on.
Figure 5.5 compares the trend of this line with a fixed fractional method,
always using 5%.
Note that the result is almost double, but the drops during the drawdown
periods are worse, too. Note also that, in fact, although a higher percentage
is used on anything over the initial capital (or multiples of the same), the
Timid Bold 5%-23.4% Vs Fixed Fractional
1,600,000.00
1,400,000.00
1,200,000.00 1,206,210.00
Timid Bold
1,000,000.00 %f 5%
800,000.00
654,917.50
600,000.00
400,000.00
200,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.5 A comparison between a standard line and one that risks more on
accumulated gains.
final result represents a higher percentage for all the capital. Starting with
three contracts instead of two is simply more risky. It doesn’t matter if the
102 third contract is used applying an imaginative formula; you’re simply risking
more. To prove this, we could apply the same concept to a base percentage
of 4% and 23.4% to anything over the initial capital. The results are shown
in Figure 5.6.
Note how the fixed 4% line proceeds as usual without a hitch (remember
3.75 was the minimum that could be used for the series of trades using this
system), while the timid/bold line is blocked.
This is because more was risked at the beginning and the initial negative
period brought the line to €30,997.5, which wasn’t enough for even one
contract:
( )
(30, 997.5 ∗ 4%)
contracts = ENT = ENT (0.99192) = 0.
1,250
Risking always and only 4% the equity never dropped this low so the
system could continue to trade.
This is an example of how a more aggressive strategy can produce both
damage and benefits. What’s important is to know this, and therefore be
prepared for the negative consequences such a choice might entail.
Timid Bold 4%-23.4% Vs Fixed Fractional 4%
400,000.00
315,070.00
Timid Bold
%f 4%
200,000.00
30,997.50
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.6 Reducing the base percentage from 5% to 4% blocks the mixed curve.
■ 5.6 Equity Curve Trading

One concept that’s very popular with traders is adapting your trading to
103
the principal equity, the original equity with just one contract. This means

trying to stay in the saddle as far as possible, making the most of good times
and leaving it be during the bad.
One method is to calculate a moving average of the equity and check when
the two lines (the equity line and the average equity line) cross in order to
find, more or less, favourable or unfavourable trend periods. The number of
periods to use to calculate the moving average depends to a great extent on
the type of system, although often traders use 30 periods of good results.
Figure 5.7 shows the trend of the principal equity of the system and the
average of the same over 30 periods.
What this method does is prevent trading when the equity curve is below
its moving average.
Figure 5.8 shows a comparison between the lines produced by a 5% risk
using the fixed fractional method. The thick line shows that obtained block-
ing the system when the principal equity was below its average over 30
periods.
Note that the final result of this method, ceasing trading in negative times,
is worse than the complete method but, without a doubt, the trend of the
200,000.00
138,082.50
137,901.25
30 periods moving average

Equity with 1 contract
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.7 Trend of the principal equity and the crossover with the moving average
of the same over 30 periods.
%f 5% with or without stop

800,000.00
654,917.50
600,000.00
104
513,565.00
Stop equity on
400,000.00 drawdown
%f 5%
200,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
1
FIGURE 5.8 Trend of the equity with %f = 5%, blocking some trades.
equity is more ‘benevolent.’ The drawdowns are much more acceptable,

and the worst period of the system is kept ‘frozen’ to start trading again
only during the subsequent positive recovery.
Let’s take a look at what would have happened if we used the system in a
more aggressive way, respectively, with risk percentages of 7.5%, 10%, and
finally, 15%, which, as you may recall, was a disaster.
%f 7.5% with or without stop
2,000,000.00
1,549,707.50
1,800,000.00
1,600,000.00 1,546,612.50
1,400,000.00
1,200,000.00
Equity stop on
1,000,000.00
drawdown
800,000.00 %f 7.5%
600,000.00
400,000.00
200,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.9 A %f of 7.5% produces results that are practically the same for the two
methods.

4,000,000.00
3,800,000.00
3,600,000.00
3,400,000.00
3,200,000.00
3,000,000.00 105
2,800,000.00 2,769,745.00

2,600,000.00 1,967,672.50
2,400,000.00
2,200,000.00
2,000,000.00
1,800,000.00 Equity stop on
1,600,000.00 drawdown
1,400,000.00 %f 10%
1,200,000.00
1,000,000.00
800,000.00
600,000.00
400,000.00
200,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.10 The method that doesn’t trade in the downturns ends up producing
much better results with an aggressive %f.
It’s worth taking the time to reflect on what we can see in this line. It
isn’t an invitation to increase risk percentages with no limits, ceasing trad-
ing when the system goes into a stall; but it does show how a system one
might call ‘irregular’ with nasty drawdown periods, as every moving aver-
age crossover system tends to be, can be improved by adapting your trading
10,000,000.00
9,000,000.00
8,000,000.00
7,000,000.00
6,000,000.00
5,273,425.00
5,000,000.00
Equity stop on
4,000,000.00
drawdown
%f 15%
3,000,000.00
2,000,000.00
7,822.50
1,000,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.11 Ceasing trading at the worst times can produce impressive results.
to the trend of the system. Note that, for a moving average crossover sys-
tem more than others, the trend reflects the market period and, adjusting to
the results of the system means adjusting to the market’s behaviour in that
particular period.
106 This is what happens using the fixed fractional method, ceasing trading
below the average.
delta = 5,000
800,000.00
600,000.00
582,200.00
508,892.50
Fixed ratio with

400,000.00
stop on drawdown
Fixed ratio
200,000.00
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.12 With stops below average the fixed ratio equity improves.
Now let’s take a look at the fixed ratio methods using the same technique,
in other words making no trades in the periods in which the equity with 1
contract is below the average of that equity over 30 periods. We’ll consider
cases with a delta of €5,000, €2,500 and €1,250 respectively; in other words
starting with a ‘normal’ delta and getting more aggressive.
delta = 2,500
1,200,000.00
1,000,000.00 969,642.50
869,507.50
800,000.00
600,000.00 Fixed ratio with

stop on drawdown
Fixed ratio
400,000.00
200,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
107
FIGURE 5.13 Fewer drawdowns with an aggressive delta and stops below the moving

average.
delta = 1,250
1,800,000.00
1,600,000.00 1,577,102.50
1,400,000.00
1,303,430.00
1,200,000.00
1,000,000.00
Fixed ratio with
800,000.00 stop on drawdown
Fixed ratio
600,000.00
400,000.00
200,000.00
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.14 A very aggressive delta, which produces good results with acceptable
fluctuations.
%vol 3%
400,000.00
240,485.00
234,935.00
200,000.00 %vol with stop

on drawdown
%vol standard
0.00
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
81
1
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
FIGURE 5.15 %vol of 3%; the results are the same but the line is more regular with
stops below the average.
To conclude this analysis, let’s take two cases applying the percent volatil-
ity method with an allowed volatility of 3% and 15%, the first quite a delicate
approach, the second very aggressive.
108 Note that, for lower values of the allowed percent volatility, the result
doesn’t change a great deal. The only difference is that there are some
more easygoing periods at the times when the market was unfavourable
for the strategy. You’ll recall that at low %vol values the system itself
prevents trading at various entry points, automatically filtering potentially
hazardous ones.
The scenario with a very aggressive percentage, such as an allowed volatil-
ity of 15%, is very different. The final result, ceasing trading at times when
the principal equity is below the 30-day average, produces almost 10 times
what we would have obtained if we continued trading. A look at Figure 5.16
shows this is mainly due to keeping equity high after the seven-hundredth
trade.
Obviously, while on the one hand the final result is certainly remarkable,
the drawdown suffered from the highest equity levels is also quite painful.
In fact, I believe this example should only be taken as mathematical proof
of the method, and I wouldn’t want to encourage anyone to use such levels
of risk.
%vol 15%
10,000,000.00
9,000,000.00
8,000,000.00
7,000,000.00
6,000,000.00
5,000,000.00
4,421,800.00
4,000,000.00
%vol with stop
3,000,000.00 on drawdown
%vol standard
2,000,000.00
1,000,000.00
468,662.50
0.00
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
42
83
1
12
16
20
24
28
32
37
41
45
49
53
57
61
65
69
73
78
82
86
90
94
FIGURE 5.16 An extreme %vol, not trading at unfavourable times, changes the
results considerably.
It must be said that blocking the system below the moving average
improves the final situation for systems that suffer marked drawdown
periods. Ceasing trading, in fact, more often than not, prevents following
the signals during prolonged depreciations in the equity. Often, the very 109
configuration of a system alternates good and bad periods, and this method

aims to cease trading in the bad ones. Remember, we’re working with
a banal moving average crossover system, and it’s well known that these
systems work well in trend periods, while suffering heavy drawdowns when
the market becomes congested. It’s also well understood that the market
alternates trend periods with congested periods, and by ceasing trading
below the equity’s moving average we can often stay away from long periods
of congestion.
That’s not all. This type of stop also protects you from the risk of finding
yourself with a system that, for whatever reason, no longer works. If the
system is doing badly, this acts as a sort of emergency brake applied before
you lose the profits you managed to accumulate up to that point.
There are other systems, though, that – thanks to special filters – produce
a much more regular equity line with much less marked drawdowns. If
we tried to apply a system stop of this kind to one of these systems, we’d
always and inevitably obtain worse historical results. We should consider,
however, that these systems already have a filter upstream that cleans up the
equity trend. (It would be very difficult for a system without special filters,
intended as additional conditions, to produce a regular equity line.) Apply-
ing this filter is a good idea for the purpose of protecting yourself from the
possible ‘demise’ of the system.
■ 5.7 z-Score
Another trading technique based on past system data analyzes consecutive
trade sequences, attempting to establish whether the result of a trade (in
terms of a positive or negative result) is dependent (or not) on previous
trades.
If there was such a thing as a system that always alternated between
gain/loss anyone, after a gain, would halt trading at the following signal
and wait for the next one. This is obviously an ideal situation that’s all
but impossible to encounter, but we can study the correlations to provide
trading guidelines. There’s room in this subject for digression if we get into
statistics. In the same way as in a coin toss, there’s a 50% probability of each
toss coming up ‘heads,’ also in trading if a system has a winning percentage
110 of 54% (to choose any old percentage) every trade should have a 54%
chance of winning. The above would make the whole subject of dependence
of trades, invalid; but we can consider that, as trades derive from market
behaviour, said behaviour may crop up again continuously in trades of the
same characteristics. In other words, the event isn’t merely a question of
statistics related to one particular case, but is driven by external factors.
These are lines of reasoning everyone is free to consider, and decide
whether to get your teeth into the subject (or not). Personally, I’m not a
firm believer in the effectiveness of the same, but I do think they’re worthy
of consideration.
The z-score is used to measure how much dependence there is between
trades, with the following equation:
N ∗ (R − 0.5) − X
z-score = √
X ∗ (X − N)
N−1
In which N is the number of trades, R the number of series (a sequence
of positive trades is one series, a sequence of negative trades is another) and
X = 2 * W * L in which W is the number of winning trades and L is the
number of losing or null trades.
Positive z-scores show a tendency for alternating trades, so after a win
you could expect a loss, and vice versa. Negative scores, on the other hand,
show a tendency of strings of consecutive trades producing the same result.
In fact, the higher R is, the more alternation there is in the positive/negative
result of the trades and the higher the equation numerator will be (if there
was perfect win/loss/win/loss alternation R would be equal to the number
of trades N). If R decreases on the other hand (if all the trades were somehow
positive, R would be equal to one as there is just one sequence of trades),
the numerator would become negative and decrease more and more.
But it’s not enough for the calculated score to be positive or negative to
take decisions. There must be considerable presumed dependence to give
us a real advantage and, the higher the z-score is in terms of an absolute
value, the greater the probability the dependence found will occur again in
the future.
In particular, 1.96 (or –1.96 if negative) is a value that’s often used as a
limit value. This value gives weight to the significance of the information.
In particular, 1.96 means the information is 95% valid, which is obviously a
good value.
111
The z-score is similar to standard deviations in a Gaussian distribution. In

said distribution, 95% of the data comes within a wide range from –2 to +2
standard deviations (–1.96 +1.96 to be precise, and again we’ve found our
limit number).
If the z-score is 3 or –3, the level of confidence would be 99.73%.
If we obtained a z-score of over 1.96 for a system we might risk consider-
ing that after a winning trade it would be highly probable there’d be a losing
trade, and after two winning trades it would be even more probable there’d
be a losing trade. We could analyze the system report to check how many
series of single trades there are, and how many trades in sequences of two,
and trade as a consequence, ceasing after one or two winning trades, until
there’s a losing trade (obviously only a signal; as we’d be faithfully following
the z-score).
Vice versa with a negative z-score of less than –1.96, we could cease trad-
ing after the first losing trade, confident that it would be followed by other,
numerous, losing trades; and only start trading again after the first winning
trade. With a negative z-score in fact, the trades should tend to maintain the
same sign for quite long sequences.
Unfortunately, the system we’ve been using for all these examples, isn’t
well-suited to this theory. Let’s calculate the z-score.
An analysis of the sequence of the 964 trades produces the following data:
N = 964
R = 486 (243 winning sequences and 243 losing sequences)
W = 473
L = 491
X = 2 ∗ W ∗ L = 2 ∗ 473 ∗ 491 = 464,486
964 ∗ (486 − 0.5) − 464,486
z-score = √ = 0.2364
464,486 ∗ (464, 486 − 964)
964 − 1
As can be seen, the result is extremely low, too low to base any sort of
specific strategy on.
Apart from the fact that this method can’t really be effectively applied to
the example we’ve been using, it must be said that, usually the more trades
that are used to calculate the z-score, the more this method can be considered
valid. In our case, 964 is a significant number; but doing the same thing on
112 fewer than 300 trades wouldn’t produce very reliable results. We could, for
example, calculate the z-score trade by trade, and see how it can change
considerably, and therefore be quite untrustworthy.
■ 5.8 Conclusions
Once you know the basic methods that can be used to apply money man-
agement strategies to your trading, it’s also important to know these can be
tweaked in various ways. A trader should have a clear idea of his goal, max-
imum risk or maximum results, limited drawdown, low exposure always,
and in any case, etc. Once you know your goals, you can adapt the money
management method to your own risk profile.
We’ve looked at various ways to make an approach more aggressive in the
initial stages of the system, and make it more docile at later stages. Other
methods go in the opposite direction, attempting to risk more as profits are
added to the system.
Some methods try to follow market trends or ‘standby’ when they are
unfavourable for the system.
Finally, we considered an instrument that can be used to try and define
the predictability of the next trade and decide whether to follow it or wait
for the next one.
The possibilities of studying new techniques and combining those men-
tioned above offer a multitude of scenarios, so even the most demanding of
traders can calculate the optimal number of contracts to use.
■ References
Jones, Ryan. The Trading Game: Playing by the Numbers to Make Millions. New
York: John Wiley & Sons, 1999.
2007.
113

CHAPTER 6
The Monte Carlo

Simulation
At this point, you could already choose the money management method you
want to use. In fact, you already have enough information to roll your sleeves
up and get to work, after deciding how you want to use your strategy.
114 But I must once again dampen the enthusiasm of those who, having seen
the firepower a money management strategy can bring to bear, can’t wait
to put the pedal to the metal and start trading with high-risk percentages
using the fixed fractional method, or a very low delta using the fixed ratio
method. In the previous chapter, I always emphasized that there’s another
side of the coin, which, apart from drawdowns that can be difficult for any-
one to withstand, can also put an end to your strategy if your capital is now
insufficient.
■ 6.1 Using the Monte Carlo Simulation

Let’s take a look at an example. Supposing a trader has €20,000 and uses a
strategy with a stop-loss of €1,500. On the basis of an analysis of his strategy’s
historical data, he decides on an extremely aggressive approach (the only one
possible with a low initial capital) using the fixed fractional method with a
10% risk.
Let’s suppose his sequence of trades is as shown in Table 6.1.
TABLE 6.1
Capital No. of contracts Trade
20000 1 500
20500 1 600
21100 1 450
21550 1 800
22350 1 −3000
19350 1 −2100
17250 1 10000
27250 1 2750
30000 2 1000
32000 2 ...
There are two losing trades, and both lose more than the theoretical
stop-loss (one even twice as much). We’ve seen, and I’ve said several times,
that this isn’t something that can be excluded and, on the contrary, it’s some-
thing that happens more often than one might imagine.
The final result, though, is quite respectable and paves the way for the
trader to proceed with his work. 115
THE MONTE CARLO SIMULATION

If this series was the series of trades obtained in the back testing, the trader
would have been able to rely on it and choose a risk percentage of 10%,
considering the 50% gain (from €20,000 to €30,000) at the end of the series.
But let’s see what would have happened if the order of the trades was
simply different to that in Table 6.1, and was as shown in Table 6.2, for
example.
As you can see, the final result is very different to the previous one. While
still making a discrete profit, the trader is certainly in a worse condition than
he thought he’d be in.
But that’s not all; Table 6.3 shows another case with a different distribu-
tion of trades.
What happened? It’s simple. With just €14,900 remaining after the first
two losing trades, the trader no longer has sufficient capital to continue using
a studied strategy; in fact, 10% of €14,900 is 1,490, which isn’t enough to
cover the system stop-loss of €1,500.
Obviously, the trader could increase his risk percentage and continue,
but this stratagem should be avoided, as it’s better to stop using a strategy
TABLE 6.2
20000 1 10000
30000 2 −3000
24000 1 2750
26750 1 800
27550 1 500
28050 1 600
28650 1 450
29100 1 1000
30100 2 −2100
25900 2 ...
TABLE 6.3
capital No. of contracts Trade
20000 1 −300
17000 1 −2100
116 14900 0 450
14900 0 800
14900 0 500
14900 0 600
14900 0 10000
14900 0 1000
14900 0 2750
14900 0 ...
that’s already put you in a condition from which you can’t continue. In this
case, there would have been an excellent recovery, but the future is always
an unknown factor and the market might not return what you’ve just put
into it.
This is an example of how the fortuitous nature of events can change the
final scenario considerably. If the trader used just one contract, those trades
would always have produced the same final result anyway, but that wouldn’t
be the case if he used a money management strategy. If those trades are a
reflection of the market on which they were made, there’s nothing to say
they won’t occur in exactly that way, but in a totally different order to what
was predicted.
If, instead of 10%, the same trader risked 11% of his capital, with the
same sequence in Table 6.3, he would have obtained the result in Table 6.4.
TABLE 6.4
20000 1 −3000
17000 1 −2100
14900 1 450
15350 1 800
16150 1 500
16650 1 600
17250 1 10000
27250 1 1000
28250 2 2750
33750 0 ...
117
This is a case in which a greater risk than what is already considered

aggressive, produces completely different results. This doesn’t mean you
should risk more, but it does mean that apparently insignificant details like
the order in which the trades occur can produce substantial differences. Also
note that the above example is totally fictitious for the sole purpose of illus-
trating these differences on the basis of the order of events. What increases
the risk from 10% to 11% could also be considered finding a minimum risk
percentage to continue without worrying about the order of the trades. In
the real example in the previous chapter, the minimum percentage for the
fixed fractional method was 3.75%, below which the system ceased trading.
So, is there a way to protect yourself from such hazards?
There is certainly not a definitive way, above all, because the market can
change radically, but we can at least evaluate the effect of various sequences
of trades using simulations, and try to choose the best approach.
This is why a version of the Monte Carlo simulation is used in trading.
The Monte Carlo simulation is a method that, on the basis of some of the
data at your disposal, you attempt to extrapolate other data to produce a
more varied series of cases. In practice, you analyze the known sample, and
by studying the statistical characteristics you ‘prepare’ other data with the
same characteristics.
One quite well known application of the Monte Carlo simulation was
during World War II, when it was used to try and optimize bombing raids.
Another classic application of this method is calculating how long a person
will wait at a bus stop on the basis of how often the buses pass, the number
of people travelling, etc.
The Monte Carlo simulation fell on fertile ground in the trading world,
and it’s used in various ways.
One of the most popular methods consists of taking a series of trades you
have at your disposal and using them to create a high number of permutations
of the same; in other words, creating new series of the same trades, but in
different positions.
Let’s take a look at an example of the permutation of three elements: A,
B, and C. We could end up with the following cases:
TABLE 6.5
A A B B C C
B C A C B A
118
C B C A A B
The number of permutations of n numbers is n! (factorial n), which is

equal to n*(n − 1)*(n − 2)* . . . *1, in our case 3*2*1 = 6, exactly the
number of cases found.
In general, the approach simulates 1,000 or more new series of trades and
applies the chosen money management strategy to each of these. This creates
a statistical sample of results, and we can see how they change. The greater the
distribution of the final profits, or the maximum percentage drawdowns, the
less reliable the choice made is, which clearly depends a lot on how the events
occur, and a change in the expected order can make a notable difference (for
better or naturally for worse).
This approach gives you an idea of how things can change compared to
what you experienced while following pretty much the same pattern as the
one you’ve just seen.
A more complex method lets you reutilize the trades you’ve already used
in the new series. In practice, it doesn’t create permutation of the principal
series, but creates new series picking trades randomly from a list. A trade
might appear several times in the new series, and therefore its effect would
be multiplied. If this repeatedly used trade is by chance the worst-ever loss
of the system it would have a pessimistic effect, in favour of safety.
We’ll use this simulation system in our examples, creating a certain num-
ber of new lists of trades picked randomly from the original list of our trading
system.
There are also even more complex methods that, for example, let you
forcibly add the maximum historical loss of the original system a certain
number of times, or randomly add a value that’s double the same. This would
simulate a day on which a particular event produces a notable crash (never a
good thing for traders, who almost always find themselves flat on their faces
when something unexpected happens).
A much more complex method goes even further and studies the history
of every trade. In particular, it examines how the market has moved since
the trade opened until when it closed, for the entire duration of the trade,
recording the percentage fluctuations of each bar, and remixing these fluc-
tuations at random, it creates a new history for the trade, changing the final
result in fact. In practical terms, let’s say this system creates new series of
trades in which the individual results are different to the original ones. The
only thing they have in common is a certain harmony with the market you
119
are trading on. This is certainly a very sophisticated method, but in my opin-

ion it goes beyond the spirit of a simulation and almost leaves the world of
trading behind to become an exercise in statistics. Furthermore, in order
to apply a method of this kind, you’d need to use tools more sophisticated
than I can imagine, and the amount of information necessary would be con-
siderable too. A good software tool for this and other methods is Monte
Carlo-Lab, to be used with Wealth-Lab, but I must warn you that it can
be quite complicated to program strategies in Wealth-Lab, and use Monte
Carlo-Lab.
If you want to create new series of trades you can start also on the basis
of the statistical distribution of the original trades, using the average and the
standard deviation, to create new values using these parameters.
I’ve included a program in Excel with the book you can use to run money
management strategy simulations with the fixed fractional and fixed ratio
methods.
Enter the list of trades and principal data in the spreadsheet to run the
simulation, in particular the number of available trades, the type of money
management method you want to use in the simulation, the risk profile (the
risk percentage if you choose the fixed fractional method, or delta if you
choose the fixed ratio method) and, just as important, the number of simu-
lations you want to run.
As a number of reference, it’s a good idea to run at least 1,000 simula-
tions, 5,000 will give you a more reliable result, and even 10,000 will make
the results more trustworthy, although there’s not that much improvement
to make the longer computing times worth it.
The other data to enter includes the margin per contract (to limit the
number of contracts when the available capital isn’t sufficient to cover
the margins), and the maximum number of contracts you want to work
with. This number is a limit put on trading to avoid disrupting the market.
If you’re lucky and find your capital has increased considerably, you might
find yourself in a situation in which, when calculating the number of con-
tracts to use, the result is a very high number. If, for example, the calculation
suggests trading with 100 SPMIB futures, it’s obvious that this number of
contracts will be difficult to manage, not because you’ll cause problems for
other traders on the market (100 new contracts wouldn’t upset the market)
but it will be problematic to enter trades with 100 contracts at the desired
price in a reasonable amount of time. 100 is in any case just an example,
and you could even find yourself faced with higher numbers, which would
120
be impossible to implement, so the maximum number of contracts is a
limit you set in order to be able to effectively trade on the market.

The program extrapolates the maximum historical loss on the basis of
the list of trades; you can choose to use this figure as the maximum loss, or
continue using the theoretical stop-loss.
It’s definitely better to use the maximum historical loss if you want to
protect yourself from similar events in the future. In any case, the examples
in the previous chapters were all concluded using a theoretical stop-loss.
Later, we’ll see what difference choosing one parameter rather than another
can make.
Apart from entering the initial capital as a start point, there’s another
important figure to enter, the number of trades to simulate. For the basic
purpose of our work, this number could be the same as the historical trades.
Later on, we’ll go into how this could be used in a different way.
The first simulation we’ll study applies the fixed fractional method with
a 5% risk percentage and a maximum loss equal to the €1,250 stop-loss. An
initial capital of €50,000 is used, in line with the examples in the previous
chapters.
The first chart to analyze is the distribution of the percentage returns.
A histogram was created in which each column shows the number of cases
with a percentage return of that level.
Figure 6.1 shows the histogram.
The histogram provides a great deal of information. The first thing that’s
evident is there’s a 50% probability of making more than 1,000% over a
period of roughly six years (the 964 available trades represent six years using
our system). Note, in fact, that 476 of the 1,000 simulations closed above
1,000% (for practical purposes, results over 1,000% are always grouped
together).
This information might already be sufficient for a trader to go to work
with the much-celebrated 5%.
Figure 6.1, however, tells another story. For example, we can see that the
distribution of the cases is very ‘flat.’ The peak over 1,000% is the result of
grouping together all the cases with percentages higher than this figure and,
if we created other percentage classes above this figure, we’d probably end up
with an even flatter histogram. Some might say there’s a significant column
also before the last with 175 cases. The truth is that this column shows the
cases with returns from 500% to 1,000%, which is quite an extensive range.
All the other ranges are 50 percentage points from 200% up (200%, 250%,
121
300%, etc.) and the range is 10 points for lower values.

500
476
450
400
350
300
250
200
175
150
100
50
25 28 29 27 30 24 20
1 7 3 3 7 5 9 14 10 7 8 5 7 8 8 8 12 6 7 5 6 6 13
0
m 0 t 50
m 0 t 40
m 0 t 30
fro 20 –20
fro 10 10
fro 0 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 80 80
fro 90 o 90
m 0t 0
m 0 to 10
m 0t 0
m 0 t 30
m 0 t 40
m 0 t 50
m 0 t 60
m 0 t 70
m 0 t 80
m 0 t 90
m 0 t 00
m 0 t 50
m 0 t 00
m 0 t 50
m 0 00
m 0 50
to 0
> 00
00
fro 0 10
fro 12 12
0 50
m to
fro 11 o 1
– –
fro 13 o 1
fro 14 o 1
fro 15 o 1
fro 16 o 1
fro 17 o 1
fro 18 o 1
fro 19 o 1
fro –5 <–
fro –4 o –
fro –3 o –
fro 20 o 2
fro 25 o 2
fro 30 o 3
fro 35 o 3
fro 40 o 4
fro 45 to 4
10
10
m to
m to
m to
m to
m to
m to
m to
m to
m to
m to
t
– o
50 to
1
m
m
fro
FIGURE 6.1 Distribution of percentage returns with a 5% risk.

A distribution of this kind substantially indicates that a 5% risk creates
an outcome that isn’t easy to predict. We know that in 50% of the cases,
returns will be over 1,000%, but above this percentage or in the other half
of the histogram it’s impossible to predict the final result in a reliable way
(in other words, you can’t confidently aim for 400%, for example).
We’ve considered the positive aspects of the histogram, boldly going over
1,000%, but it’s important to consider the left part too, where various sim-
ulations produced negative results.
In particular, 25 cases left less than half the capital intact, and in another
11 cases the final capital was less than the initial capital (three with losses of
up to 10%, seven from –10% to –20%, and one from –30% to –20%).
If we add all the negative cases, in total there are 36 cases in which the
system is a losing one: 36 out of 1,000 is 3.6%, which means that using a
5% risk, there’s a 3.6% probability of the result being a loss after six years.
We’ve considered the returns and their range of percentage values, but
the software provides other information that’s just as important when assess-
ing the approach: the maximum percentage drawdown.
Figure 6.2 shows the chart with the distribution of this parameter.
The distribution is much more regular than that of the percentage returns.
At a glance, we can see the most populated zone around 49%, but we can
122
also say without a doubt that most of the results are between a 34% and a
70% percentage drawdown.
DD distribution
60
50
40
30
20
10
0
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
FIGURE 6.2 Drawdown values for the 1,000 simulations with a 5% risk.
So, trading with a 5% risk and a maximum loss equal to the system
stop-loss, there’d be a 50% probability of making over 1,000% but
with a drawdown that would most probably be between 35% and 70%.
Considering the other aspects of these results it’s now up to the trader to
decide whether this is acceptable or not. For example, is the trader willing
to have an almost 4% probability of losing money after trading for six years?
Let’s not forget in any case that, apart from the cases in which there’s a loss,
there are various cases in which the percentage gain in six years might not
be considered sufficient for some traders.
Now let’s take a step back and consider the number of simulations again.
I said 1,000 was a good number, but you could run up to 5,000, for more
precise and reliable results. Out of curiosity, let’s see what the histogram
would look like with 5,000 simulations.
Figure 6.3 shows more or less the same situation as Figure 6.1, with 2,394
cases out of 5,000 that make over 1,000% and 210 out of 5,000 that pro-
duce a loss. The cases that produce high returns are still just less than 50%
while 210 out of 5,000 is 4.2%, which is quite close to the 3.6% calculated
previously.
Figure 6.4 shows the drawdown.
123

3000
2394
2500
2000
1500
916
1000
500
145 153 153 157131 124 123

2 5 21 37 17 14 18 34 31 24 23 36 58 47 29 25 30 38 32 34 39 34 37 37
0
0
m 0 t 40
m 0 t 30
m to 0
m t 0
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
to 0
0
m 0 t 10
m 0 20
m 0 30
m 0 40
m 0 50
m 0 60
m 0 70
m 0 80
m 0 90
m 0 00
m 0 50
m 0 00
m 0 50
fro 45 to 0
m 0 450
to 0
> 0
00
fro –10 –1
fro –5 <–5
fro 20 –2
fro 80 8
fro 90 o 9
fro 10 10
m 0 0
0 50
0
fro 11 to 1
fro 12 o 1
fro 13 to 1
fro 14 to 1
fro 15 to 1
fro 16 to 1
fro 17 to 1
fro 18 to 1
fro 19 to 1
fro –4 to –
fro –3 o –
fro 20 to 2
fro 25 to 2
fro 30 to 3
fro 35 to 3
fro 40 to 4
10
10
m to
m to
m to
m to
m to
m to
m to
m to
t
– o
50 to
m 0
m 0
m
m
m
fro
FIGURE 6.3 Distribution of returns with a 5% risk in 5,000 simulations.

DD distribution
250
200
150
100
50
0
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
FIGURE 6.4 Trend of the percentage drawdown with a 5% risk in 5,000 simulations.
Also in this case, there is an evident similarity with Figure 6.2. The curve
shows the maximum population density in the zone around 50%. Compared
to the previous histogram, though, it looks like there are slightly more cases
with a drawdown of around 50%, but below this value there are more bars
124
from 35% to 50% than in the previous case.
So you can obtain useful information from 1,000 simulations, and if you
wish, when close to taking your final decision, you could run 5,000 simula-
tions to create a more detailed histogram.
Let’s proceed with some additional analysis to see what sorts of possibil-
ities might be presented and how to face them. We’ll look at this step by
step, just as if we were making a trading decision.
Let’s suppose you’re a particularly conservative and prudent trader, who
absolutely doesn’t want to go too far with the risk percentages. On the basis
of the analyses in the previous chapters we saw that, below 3.75%, the sys-
tem would inevitably cease trading. But this belongs to the past, and what
we want to do now is look at what might happen if we risk very little.
The stop-loss is 1,250, which is 2.5% of the initial capital. Obviously, if
we try to use a 2.5% risk percentage, all it would take would be one first
losing trade to block the system. In fact, we’d end up with a capital of less
than €50,000, and 2.5% of the new capital would in any case be less than
the estimated loss of 1,250, making it impossible to continue.
Being an overly prudent trader you might in any case test a 2.5% risk, per-
haps with the option of using a higher percentage if the system was blocked
immediately.
Figure 6.5 shows the equity trend.
Well, the histogram leaves little room for doubt, in 89.5% of the cases
you’d end your adventure with a loss of up to 10%. In actual fact, these losses
would be closer to 0 than –10%; but, in any case, they’re sufficient to cease
trading. There’s nothing to say these are all losses on the first trade, all it
takes is one trade that reduces the equity to less than €50,000 and your run
is over.
Note that there are very few cases in which the system really took off, and
of these, 18 (1.8%) produce returns over 500%.
No matter how prudent you are, you probably wouldn’t want to start
trading in these conditions, and it’s no use looking at the drawdown chart as
we already know that, with a risk of just 2.5% there’s very little chance of
continuing to the end.
The real system needed at least 3.75% to function, but if we want to try
something more conservative, let’s try 3%.
Figure 6.6 shows what might happen.
125

1000
895
900
800
700
600
500
400
300
200
100
2 1 1 4 1 2 4 1 2 1 1 5 4 8 11 12 9 10 5 3 17 1
0
m 0 50
m 0 40
m 0 t 30
fro 0 t 20
fro 10 10
fro m 0 to 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
m to 0
fro 70 70
0
t 0
m t 00
m 0 t 20
m 0 t 30
m 0 40
m 0 50
60
m 0 t 70
m 0 180
m 0 t 90
m 0 00
m 0 50
m 0 00
m 0 50
fro 4 to 0
50 to 0
to 0
> 00
00
m 0 t 10
fro 60 o 6
fro m 8 to 8
m 0 9
m 0 0
m 50 45
0 50
m o–
fro 100 o 1
fro 12 o 1
fro 13 o 1
fro 14 o 1
fro 15 o 1
fro 16 to 1
fro 17 to 1
fro 19 to 1
fro –5 <–
fro –4 to –
fro –3 to –
–2 o –
fro 20 o 2
fro 25 to 2
fro 30 o 3
fro 35 to 3
fro 40 to 4
10
10
m to
fro 11 o 1
m to
m to
m to
m to
fro m 9 0 to
fro 18 o
m t
t
–
m 0
m
fro
FIGURE 6.5 Trend of returns risking the minimum that’s mathematically possible
equal to 2.5%.
250
214
200
185
150
100
79 76
47 51 47
50
39
31
17 19 17 14 20 21 17 19
16
8 9 8 11
5 5 7 5 7
2 2 2
0
0
m 40 40
to 0
fro 0 t 0
fro 10 0
fro 0 o 0
fro 1 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 80 80
m 0 t 90
t 0
0
m 20 20
t 30
m 0t 0
to 0
m 0t 0
m 0 t 70
to 0
0
m 0 t 00
to 0
0
m 50 50
fro m 4 to 0
50 to 0
to 00
> 0
00
– 1
fro 11 o 11
fro –5 <–5
0 –3
–2 –2
fro 14 o 14
5
fro 16 16
fro 180 18
fro 19 19
fro 50 25
fro 30 30
0
m 50 45
0
m o–
fro 100 o 1
fro 1 to 1
fro 130 to 1
fro 150 o 1
fro 7 1
fro – to –
fro 20 to 2
fro 3 to 3
fro 00 4
0 5
10
10
m to
m t
o
m to
m to
m to
m to
m to
m to
fro m 9 to
o
o
m 0t
to
t o
m 0
m 0
m 0
m 0
3
1
–
4
m
m
m
m
m
m
fro
fro
FIGURE 6.6 Trend of returns for 1,000 simulations with a 3% risk.
The scenario has changed considerably compared to the previous test and
now there are 219 cases out of 1,000 that close at a loss (probably blocked
before the end due to insufficient capital) but the rest make a profit and most
126
of the cases are in the high percentage zone.
Would you be willing to risk being blocked in 21.9% of the cases (219
out of 1,000) with a probability of 26.1% (185 + 76 cases out of 1,000) of
making over 500%?
It’s worth thinking about that question before answering yes or no, and
to do so let’s take a look at the percentage drawdowns.
As can be seen, most of the drawdowns are between 15% and 40%,
with the highest probability around 25%. Frankly, this trend leaves room
for doubt; the idea was not to risk much as you know you can’t withstand
high drawdown percentages. It can’t be said that the values in Figure 6.7 are
high, but perhaps there’s something better suited to a very prudent trader.
If you remember, I mentioned the fixed ratio method appears to be better
suited to limiting drawdowns.
First, let’s run a simulation with the ‘usual’ delta of 5,000 just to see what
the scenario looks like (Figures 6.8–6.9).
Though the trend of the percentage returns is certainly interesting, the
drawdown trend isn’t. Note that just 21 cases out of 1,000 close with a
loss, while 500 close making over 1,000%. The results are similar to those
0
10
20
30
40
50
60
fro 70
0
100
200
300
400
500
600
m
fro –5 <–5 1
m 0 0
fro – to – 3
m 40 40
t 5
fro –30 o –
m to 30
–2 –2 7
6
FIGURE 6.7
fro 0 t 0
m o–
– 1
9
fro 10 0
m t 11
4 11
fro 0 o 0
m to 13
fro 1 10
21
with a delta = 5,000.

m 0t
o 15
8
fro 20 20
m to
9
fro 30 30 17
m to
8
fro 40 40 19
m to
6
fro 50 50 21
m to
fro 60 60 23
m to
4 11
fro 70 70 25
m to
4
fro 80 80 27
fro m 9 to
5
m 0 t 90 29
fro 100 o 1
0
8
m t 0 31
fro 11 o 11
0
5
m 0 33
fro 1 to 1
4
m 20 20
fro 30 11 to 35
m
6
t 30
fro 14 o 14
DD distribution
37
1
m 0t 0
fro 150 o 1
m 5 39
8
to 0
fro 16 16 41
8
m 0t 0
fro 7 1 1 o
43
2
m 0 t 70
o
fro 180 18
4
m
45
to 0
fro 19 19
m 0 0
5
47
fro 20 to 2
5
m 0 t 00 49
2 o
fro 50 25
m to 0 51
fro 30 30
m 0 0 53
fro 3 to 3
m 50 50
fro 400 to 4 55
0
fro m 4 to 0
m 50 45 57
50 to 0
0 5 59
to 00
27 33 26 23 22 22
10
0 61
> 0
193
FIGURE 6.8 Percentage returns for 1,000 simulations using the fixed ratio method
10
00 63
500
Trend of the percentage drawdowns for 1,000 simulations with a 3% risk.

127
DD distribution
50
45
40
35
30
25
20
15
10
0
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
FIGURE 6.9 Drawdown trend for 1,000 simulations using the fixed ratio method with
a delta = 5,000.
in Figure 6.1, but in terms of returns, we can consider them to be better,

as there are more high-return cases and slightly fewer cases that close with
a loss.
128
In terms of drawdowns, the situation doesn’t change much as far as the
average is concerned, which we could consider to be around 50%; but, note

that the distribution is a little wider than in Figure 6.2, although there are a
few cases of around 20% that weren’t seen with the fixed fractional method.
This could be one of those cases in which it’s worth running more simulations
to see a more precise drawdown distribution. Figure 6.10 shows the trend
with 5,000 simulations.
This histogram confirms that the drawdown risk is very similar to that
with the fixed fractional method using a 5% risk, but the bulge of bars has
moved slightly to the left, and now there is a certain (limited) number of
less-dramatic cases.
Remember, you’re a very prudent trader, who feels decidedly uneasy with
these drawdown values. Knowing you can produce less aggressive lines by
increasing delta with the fixed ratio method, you can run a test with a delta
that’s very close to the maximum historical drawdown of the original system
with one contract. Remember, this drawdown was about €22,000, so we’ll
run the test with a delta of 20,000.
Figure 6.11 shows the result.
fro
m
0
20
40
60
80
100
120
140
160
180
200
0
50
100
150
200
250
300
350
fro –5 <– 1
m 0 50
fro –4 to – 4
m 0 40
fro –3 to – 7
m 0 t 30
–2 o – 10
2
fro 0 t 20
m o– 13
FIGURE 6.11
1
fro –10 10
m 16
5
fro 0 to 0
m to 19
2
fro 10 10
m to 22
5
fro 20 20
m to 25
fro 30 30
m to
fro 40 4 28
13 15
m to 0
7
fro 50 50 31
m to
4
fro 6 60 34
m 0t
3
fro 70 o 7 37
m t 0
fro 8 o 8
m 0 0 40
fro 90 to 9
m t 0
8 13
43
fro 10 o 1
the fixed ratio method with delta = 5,000.

m 0 00 46
fro 11 to
m 0 110 49
fro 12 to 1
m 0 t 20
18 22 19
fro 13 o 1 52
m 0 30
7
fro 14 to 1 55
DD distribution
m 0 t 40
19
fro 15 o 1 58
m 0 t 50
fro 16 o 1 61
11 8
m 0 t 60
fro 17 o 1 64
m 0 t 70
fro 18 o 1 67
m 0 t 80
fro 19 o 1 70
m 0 90
fro 20 to 2 73
m 0 00
13 12 14 14
fro 25 to 2 76
m 0 t 50
81
fro 30 o 3 79
m 0 t 00
47
fro 35 o 3
m 0 50 82
72
fro 40 to 4
0 0 85
80
fro m 4 to 0
m 50 45 88
50 to 0
0 50
to 0
60 60
91
10
> 00
295
10
94
1,000 simulations using the fixed ratio method with a delta = 20,000.
00
69
FIGURE 6.10 5,000 simulations to verify the percentage drawdown distribution using

129
What’s not to like? The percentage returns are respectable, although cer-
tainly it’s not easy to do better than 1,000% with just 69 cases out of 1,000
that did so, but there are as many as 295 cases between 500% and 1,000%.
What’s more, most of the results are 200% or more.
In terms of losses, there are 8 cases out of 1,000 that close with less than
the initial capital, which is certainly an interesting result.
Satisfied with this scenario, let’s take a look at the drawdowns.
This histogram boosts your confidence further. Note that the average is
around 30%, with most of the cases below that value and very few above
50%.
Being a conservative and prudent trader, you decide without further ado
to take this route!
DD distribution
60
50
40
30
130
20
10
0
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
FIGURE 6.12 Percentage drawdowns using the fixed ratio method with a delta =
20,000.
Now let’s put ourselves in the shoes of a very aggressive but not totally
crazy/reckless trader, a trader who wants to maximize profits while
attempting to survive the market’s adversities as best he can.
Well aware of the fact that to reach ambitious goals you have to take bigger
risks, let’s immediately test the scenario with the fixed fractional method and
a 10% risk, again using the system stop-loss as the maximum loss.
This is the trend of the percentage returns in this case.
There’s no denying the good news, almost 70% of the cases close with
returns above 1,000%. It’s definitely an interesting result. The negative cases
800
676
700
600
500
400
300
200
87 88
100
15 12 11
7 7 5 2 1 7 2 3 1 3 3 2 5 4 7 3 3 2 1 2 5 5 4 4 10 9
0
m 0 50
m 0 t 40
m 0 t 30
fro 20 –20
fro –10 10
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
m 0 0
m 0 90
fro 11 to 0
m 0 t 110
t 0
m 0 t 30
m 0 40
m 0 t 50
m 0 60
m 0 t 70
m 0 t 80
m 0 t 90
m 0 00
fro 30 to 3 0
m 0 00
m 0 t 50
fro 4 to 0
50 to 0
to 0
> 00
00
fro 8 to 8
m 0 0
m 0 25
m 0 0
m 50 45
0 50
m to –
fro 10 to 1
fro 120 o 1
fro 13 o 1
fro 14 o 1
fro 15 to 1
fro 16 o 1
fro 17 to 1
fro 18 o 1
fro 19 o 1
fro –5 <–
fro –4 to –
fro –3 o –
fro 20 o 2
fro 35 o 3
fro 40 o 4
10
10
m to
m t
m to
m to
m to
m to
m to
m to
fro 9 to
– o
fro 25 to
t
m
m
m
fro
FIGURE 6.13 Distribution of percentage returns with a 10% risk.
might make you turn up your nose though, as 87 left less than half the initial
capital intact. A total of 108 were losing, and this means, at the end of the
day, you’d have just over a 10% chance of losing money. There aren’t many 131

cases between 0 and 500%. Which means, broadly speaking, you could con-
sider that, with a 10% risk, you’d either make a big profit or go bankrupt, but
the probability of making a healthy profit is seven times that of losing your
money. It is without a doubt an interesting scenario, as is also that shown in
Figure 6.1 with a 5% risk. In the previous case, though, the distribution was
less biased towards the right than in this case, meaning a lower probability
of very high returns, with most of the cases in a zone, where the returns are
‘simply’ high.
No matter how valiant and stubborn your resolve- and, above all, no mat-
ter how tempted you might be by potential profits- it’s always a good idea
to go a little further and analyze possible percentage drawdowns you might
encounter using this risk profile, as shown in Figure 6.14.
This histogram might dampen your initial enthusiasm quite brusquely.
Note that the drawdown percentages are very high, with a 10% risk. An
80% drawdown looks like a real possibility and worth thinking seriously
about to see whether you think you’ll have sufficient fortitude and faith in
the system to serenely face such events.
DD distribution
60
50
40
30
20
10
0
1
4
7
10
13
16
19
22
25
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
97
28
FIGURE 6.14 Distribution of percentage drawdowns with a 10% risk.
Considering the type of system in question (a moving-average crossover),

it’s understandable that there are high drawdowns. It’s quite normal, but
considering the type of market we’ve applied it to (a market split in half
after eliminating all the afternoon bars, which are those with the greatest
132
fluctuations), you might be worried about possible losses that are higher than
the stop-loss (bad news in the afternoon could produce a loss on the market
that’s much higher than expected). So, take a moment to study new scenarios
before making a decision.
Before testing a possible reduction of the risk, let’s try to see if we can
obtain some benefits by applying the fixed ratio method instead of the fixed
fractional method.
As we’re trading aggressively, we’ll use a delta of 2,500 and run 1,000
simulations.
Let’s take a look at the results (Figure 6.15).
It’s immediately obvious that potential profits are more limited, a little
under 60% of the cases are over 1,000%, and frankly there’s not a notable
difference between these results and those with a delta of 5,000 (Figure 6.8).
There are cases with higher profits, but not enough for there to be a real
competitive advantage in using this approach. Note that, in terms of losses
the situation is certainly more lenient than with the fixed fractional method,
but slightly worse (as could be expected) than the data in Figure 6.8 with a
delta of 5,000.
700
599
600
500
400
300
200
119
100
32 27
5 4 11 15 21 5 4 4 8 3 3 7 6 6 5 4 4 6 6 5
18 20 9 11 17 6
2 4 2
0
m 0 50
m 0 t 40
m 0 t 30
fro 0 20
fro –10 10
fro 0 o 0
fro 10 10
m to 0
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
m 0 0
m 0 t 90
m 0 00
m 0 10
m 0 t 20
m 0 t 30
m 0 40
m 0 50
m 0 60
m 0 t 70
m 0 t 80
m 0 90
m 0 00
m 0 50
m 0 00
fro 40 to 4 0
fro 4 to 0
50 to 0
to 0
> 00
00
fro 20 o 2
fro m 8 to 8
m 0 5
m 0 0
m 50 45
0 50
fro 11 to 1
m to –
fro 10 o 1
fro 12 to 1
fro 13 o 1
fro 14 o 1
fro 15 to 1
fro 16 to 1
fro 17 to 1
fro 18 o 1
fro 19 o 1
fro –5 <–
fro –4 to –
fro –3 o –
–2 o –
fro 20 to 2
fro 25 to 2
fro 30 to 3
fro 35 to 3
10
10
m to
m t
m to
m to
m to
m to
fro 9 to
m t
m
fro
FIGURE 6.15 Distribution of percentage returns using the fixed ratio method with a
delta of 2,500.
The distribution of the drawdowns is strange, with two peaks. It would

appear probable you’d suffer a drawdown of 48% to 50% or 60%, values
that are notably better than those in the fixed fractional simulation but worse 133

than those in Figure 6.9, so it’s fair to say that the advantage in terms of the
probability of high profits isn’t offset in terms of drawdown percentages,
which are worse than expected. This approach doesn’t appear to be what
we’re looking for, either.
As the drawdowns with the fixed ratio method seem far worse than the
improvement in percentage gains as delta decreases (and therefore the trader
trades in a more aggressive way), we won’t test lower delta values and we’ll
go back to using the fixed fractional method to try something else.
We know that a 5% risk won’t fully satisfy our ambition for profit; but we
also know that the percentage drawdowns with a 10% risk would be harsh.
Let’s look at a compromise and run 1,000 simulations at 7.5% (Figure 6.17).
This looks good in terms of returns. Compared to Figure 6.13, returns
over 1,000% have dropped slightly, and there’s an increase in those from
500% to 1,000%. The losing cases have been all but halved (Figure 6.18).
134
fro
0
5
10
15
20
25
30
35
40
m
0
100
200
300
400
500
600
700
fro –5 <– 1
m 0 50
39
fro –4 to – 4
method.
m 0 40
fro –3 to – 7
7
m 0 t 30
– o 10
fro 20 –2
3
m to 0
– – 13
FIGURE 6.17
FIGURE 6.16
6
fro 10 10
m t 16
fro 0 o 0
4
m to 19
2
fro 10 10
m to
fro 20 20 22
2
m to
fro 30 30 25
3
m to
fro 40 40 28
8
m to
fro 50 50 31
4
m to
7
fro 60 60 34
m to
fro 70 70
3
37
m t
fro 80 o 8
0
6
m t
40
fro 90 o 9
m
2
to 0 43
1
fro 0 10
m 0
5
0 46
fro 11 to 1
4
m 0 t 10 49
fro 12 o 1
6
m 0 20 52
fro 13 to 1
5
m 0 30 55
DD distribution
fro 14 to 1
1
m 0 40 58
fro 15 to 1
Returns with a 7.5% risk percentage.

5
m 0 t 50
fro 16 o 1 61
4
m 0 60
fro 17 to 1 64
3
m 0 70
fro 18 to 1 67
6
m 0 80
fro 19 to 1 70
5
m 0 90
fro 20 to 2 73
3
m 0 00
fro 25 to 2 76
m 0 t 50
fro 30 o 3 79
m 0 00
fro 35 to 3 82
m 0 50
fro 40 to 4
m 0 0 85
20 16 18 14
fro 45 to 0
m 0 45
8
88
50 to 0
0 50
11
to 0 91
10
> 00 94
111
Distribution of drawdowns with a delta of 2,500 using the fixed ratio
10
00 97
654
DD distribution
50
45
40
35
30
25
20
15
10
0
1
4
7
10
13
16
19
22
25
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
28
FIGURE 6.18 Trend of the drawdowns with a 7.5% risk using the fixed fractional
method.
As for the drawdowns, there is an improvement, but the values are still
quite high. Note, in fact, that there is a high probability of 60% to 80%
drawdowns.
Again, the scenario is perplexing, and the wisest decision seems to be 135
to use our resources elsewhere. This trading system isn’t what we’re look-

ing for.
As seen above, the Monte Carlo simulation can also make you take quite
drastic decisions you might not have taken, based on data from the principal
trading system alone. It’s very important to try and take a step back to ana-
lyze the results of the simulation objectively, to try and consider the possible
scenarios in terms of what your feelings might be at the time. It’s no use kid-
ding ourselves and ‘ignoring’ a 10% possibility that a certain negative event
might occur. 10% isn’t exactly a remote chance, and if you find yourself in
that situation it can be hard to continue.
■ 6.2 Maximum Loss

In the previous simulations, we deliberately used the same value as the sys-
tem stop-loss to calculate the maximum loss. We all know, both as is evident
from the figures and after due and considerable reflection, that the stop-loss
isn’t actually the maximum loss that occurred or might occur in a system.
Too many events change the rules of the game, and often the results in the
field are worse than what was expected.
In our example, we’re using a stop-loss of –€1,250; but, if we take a look
at the system performance report (Figure 3.2), you’ll see that the actual
maximum loss in the 6 years in question was €2,757.5, more than double
that!
This goes to show that, in actual fact, we’re using percentages that are
too aggressive, something that could get you into serious trouble. Risking
10%, for example, when that loss occurred you would have lost 20% of
your capital, which is certainly not something to look forward to.
Despite the fact that this maximum loss is a past event, and no one can
guarantee that a worse event might not occur in the future, it can be taken
as a point of reference for a more in-depth study, related to actual system
events. From a certain point of view, using the maximum loss instead of the
stop-loss is the same as halving the risk percentage (in this specific case, in
which the loss is roughly twice the stop-loss):
( )
C∗f
Maxloss
Doubling the value of the denominator is the same as halving the value
136 of the numerator, but using the maximum loss protects you from possible
adverse effects when rounding off.

In actual fact, studying the trend of the system, based on the real max-
imum loss, gives you a much more precise idea and follows events much
more closely.
Note also that optimal f is calculated by definition using the maximum
loss of the system.
So, let’s now take a look at what would happen to the Monte Carlo sim-
ulation if we use the maximum loss instead of the stop-loss.
A first attempt using a risk percentage of 5% is a failure and the program
doesn’t provide us with any data. Why is that? Simply because 5% of our
initial capital of €50,000 is €2,500, which doesn’t cover the maximum loss
of €2,757.5, so in this case the system can’t start trading.
Now let’s try with 10% and take a look at the results (Figure 6.19).
We can immediately see there’s been a significant change in the results.
In the previous case, in Figure 6.13, there were almost 70% of the cases
with returns of over 1,000%; this time, in Figure 6.19, they’re just over
40%. Note the scenario in terms of negative cases is particularly lenient,
and there are very few results below 0 (Figure 6.20).
0
5
10
15
20
25
30
35
40
45
50
fro
0
50
100
150
200
250
300
350
400
450
1 m <
fro –5 –5
1
m 0t 0
4 fro –4 o –
m 0 t 40
7 fro –3 o –
m 0 t 30
– o
2
10 fro 20 –20
m to
maximum loss.
FIGURE 6.20
FIGURE 6.19
– –
2
13 fro 10 10
m to
2
fro 0 0
16 m to
14
fro 10 10
19 m to
3
fro 20 20
m to
7
22 fro 30 30
m to
3
25 fro 40 40
m to
6
fro 50 50
28 m to
14
fro 60 60
m to
3
31 fro 70 70
m to
7
fro 80 80
34 m t
fro 90 o 9
4
37 m to 0 5
fro 10 10
m 0
t 0
40 fro 11 o 1
6
m 0 t 10
fro 12 o 1
7
43 m 0 t 20
fro 13 o 1
46 m 0 t 30
DD distribution
fro 14 o 1
m 0 t 40
49 fro 15 o 1
7 13 7
m 0 t 50
fro 16 o 1
6
52 m 0 t 60
fro 17 o 1
9
55 m 0 t 70
fro 18 o 1
8
m 0 t 80
58 fro 19 o 1
8
m 0 t 90
61 fro 20 o 2
8
m 0 t 00
fro 25 o 2
64 m 0 t 50
fro 30 o 3
40 36
m 0 t 00
10% risk, this time considering the real maximum loss.
67 fro 35 o 3
27
m 0 t 50
70 fro 40 o 4
m 0 00
36
fro 45 to 4
73 m 0 50
Trend of the drawdowns risking 10% with reference to the real

50 to
zones with percentages significantly lower than those in Figure 6.14.

0 50
to 0
27 27
76
10
> 00
203
79 10
00
419
In terms of the drawdowns, there’s more good news. The grouping is in
the same effect as halving the risk percentage. We have double the loss, and
We did, however, say that doubling the maximum loss produced roughly
137
by testing the system using the same risk percentages we will, in a certain
sense, be comparing apples and oranges.
As the above-mentioned aggressive trader turned up his nose at results
with a 7.5% risk, testing the same percentage now wouldn’t really provide
a concrete comparison at the same level.
It makes more sense to run the same simulation with a 15% risk, which
is twice the 7.5% that definitively blocked the aggressive trader (we’ll use
double the value as the maximum loss is approximately double the stop-loss).
This is what the trader would obtain with a 15% risk and the real maxi-
mum loss (Figures 6.22 and 6.21).
Note that the profit percentages are slightly lower, but the histogram
showing the drawdowns is slightly ‘easier to digest’ as can be seen if we
compare Figure 6.21 to Figure 6.18 (the differences aren’t great).
So, which is the best parameter to use? The choice, once again, is up to
the person using the method.
Bear in mind that, if the maximum loss diverges greatly from the stop-loss
(as in this case), you should check to see if it’s an isolated case, or whether
there have been other trades that suffered losses much higher than those
estimated. If it’s just one isolated case, the best idea is to use the stop-loss,
considering that particular loss as an unfortunate one-off; which does, how-
138
ever, in the simulations illustrate the idea of the vicissitudes one has to deal
with on the market quite well.
DD distribution
50
45
40
35
30
25
20
15
10
0
1
4
7
10
13
16
19
22
25
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
28
FIGURE 6.21 Drawdowns with a 15% risk, using the real maximum loss.
700
639
600
500
400
300
200
122
100
33
19 16 20 19 20
4 2 5 3 7 6 2 2 3 5 2 3 7 4 5 6 1 4 4 6 3 1 7 3 15
0
m 0 50
m 0 40
m 0 t 30
m to 0
fro 10 10
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
to 0
0
m 0 t 10
m 0 20
m 0 30
m 0 40
m 0 t 50
m 0 60
m 0 70
m 0 80
m 0 90
m 0 00
m 0 t 50
m 0 00
m 0 50
fro 45 to 0
50 to 0
to 0
> 00
00
fro 20 –2
fro 80 o 8
fro 90 o 9
fro 10 10
m 0 0
m 0 45
0 50
fro 11 to 1
– –
fro 12 o 1
fro 13 to 1
fro 14 to 1
fro 15 to 1
fro 16 o 1
fro 17 to 1
fro 18 to 1
fro 19 to 1
fro –5 <–
fro –4 to –
fro –3 to –
fro 20 to 2
fro 25 to 2
fro 30 o 3
fro 35 to 3
fro 40 to 4
10
10
m to
m t
m to
m to
m to
m to
m to
m to
m t
t
– o
m 0
m
m
m
fro
FIGURE 6.22 Returns with a 15% risk, using the real maximum loss.
If, on the other hand, there are many cases in which the loss was greater
than the stop-loss, it’s definitely better to use the maximum loss, especially
to calculate the number of contracts when using the system.
In fact, it all depends on what you intend to do when you start using the 139
system. If you intend to use the stop-loss in all the calculations, then you

should also use it for the simulations; the choice of the risk percentage in
fact depends on the result of the simulations, and it wouldn’t make sense to
run these with different parameters to those you intend to use.
Vice versa, if you choose to use the maximum loss to calculate the number
of contracts, the best thing is to run the simulations using the same value. In
this case, it wouldn’t be that much of a problem if you ran the simulations
using the stop-loss as the simulations were done in a more unfavourable envi-
ronment than the real one (the simulations would produce various cases with
losses greater than those expected), and using a higher maximum loss would
put us in more cautious conditions.
■ 6.3 Conclusions
Being able to simulate cases that might occur, but aren’t included in the his-
torical data, lets you analyze more cases you might find yourself in on the
basis of your money management decisions.
Rounding off to whole numbers of contracts, and increasing the num-
ber of the same as the capital increases, are important to make the strategy
you intend to use bear fruit. It’s just as important to consider the order in
which events occurred. The Monte Carlo simulation lets you create a certain
number of scenarios, based on the original, simply by mixing up the order
in which events happened; so the same might be repeated or some events
might not occur at all.
The data obtained from the simulations can be used to evaluate the prob-
ability of the results you might obtain when making a certain choice, to help
you make the right one.
140
CHAPTER 7
The Work Plan

■ 7.1 Using a Work Plan
It’s a good idea to use a strategy with a clear idea of the field you’re going
to use it in. If you simply start using a strategy without a clear idea of how
you’re going to use it, this can often put you in situations where you don’t
know what to do. As long as you’re making a healthy profit everything is fine,
but as soon as you suffer a period of drawdowns you’ll begin to worry. At 141
other times, if you know the rules of the strategy (perhaps one you created
yourself), you can try to change it to improve things in the unfavourable
period you’re in. A good trader stands out from the crowd also in these
cases due to his abilities, and one of the essential qualities of a good trader
is knowing how to react to losses.
If the trader did a preliminary study of the strategy (and therefore planned
potential goals but also, and above all, planned for potential hazards) he’d be
able to assess difficult periods more calmly, knowing whether they were part
of the game or whether he should be worried if something is effectively going
wrong.
Planning is often a job that’s given very little consideration or is done
superficially, perhaps because people just want to get started as soon as pos-
sible, or because they think it’s boring and won’t be a great help anyway. But,
when you find yourself in difficulty (and inevitably sooner or later everyone
does), if you know how to react, all the work you’ve done, and the time
you’ve put into planning up to that point, will pay off.
In this chapter, we’ll lay the foundations to give you an idea of a planning
method you can use with the Monte Carlo simulation software described in
the previous chapter. The purpose is to give you a few ideas you can base
your own work on, in the way that suits you best.
Let’s suppose we have the usual €50,000 and want to start using the by
now well-known moving average crossover strategy we discussed earlier in
the book.
In the previous chapter we looked at various ways to size a position when
entering the market, and how the equity trend changed over the six years in
question.
Planning an activity that will last for six years will certainly mean devel-
oping some interesting medium-term goals; but it doesn’t leave much room
for intermediate checks, as it’s a reflection of the situation when everything
is said and done.
There’s a parameter in the software we haven’t discussed yet, the number
of trades to simulate; in the previous chapter this number was set to be equal
to the number of available trades.
This number simply represents the length of each series of trades you’ll
create.
Let me explain: in the previous chapter we were working with 964 avail-
142
able trades, and we were picking randomly from the list 964 times (copying
THE WORK PLAN
the chosen value but keeping it, so there were always 964 trades right up to
the end); and then we created a second series of 964 trades, then a third,
a fourth and so on up to 1,000, or 5,000, or any other number. We were
creating 1,000 series as long as the original one. This was the most obvious
procedure to follow to see what would have happened to that series in the
case of a different sequence of events.
In any case, the 964 trades were those from six years of using the system,
and if we take another look at Figure 3.3 you’ll see that around 160 trades
were placed every year, a figure that can also be obtained by dividing the 964
trades by 6.
As each of the 1,000 simulations was created from the first to the 964th
trade, it’s as if each simulation was traded for those same six years. At this
point, it’s easy to see that, if instead of tabulating 964 values, we used just
the first 160, it would be as if we just studied the first year of the simulation.
If we continued to 320, it’s as if we’re simulating two years of the system’s
history, etc.
I’m sure you’ll see what I’m getting at. What I want to do is verify the
probable history after one, two, or three years (or other chosen periods of
time) to create a sort of roadmap to be able to check how close (or not) we
are to predicted performance.
Note that, by applying money management strategies, the profits (which
hopefully there are) grow exponentially, compared to the linear growth of
single contract strategies. Therefore, the scenario after one or two years
will be a lot different to that after 6. As soon as the strategy starts making a
profit, and more contracts are used, the growth of the equity soars. The flip
side of the coin is that this increase in the number of contracts also increases
drawdown percentages, which are therefore not the same when trading with
just a few contracts as when you’re trading at full steam.
The 70% drawdowns in the previous chapter shouldn’t therefore be taken
as a reference for the first year of the strategy, when one would presume they
would be lower; at the start you should run specific analyses.
If you don’t take a close look at things year by year, there’s the risk of
accepting a negative scenario that shouldn’t actually exist at that particular
time. If, for example, we started working on a strategy accepting the pos-
sibility of a 70% drawdown and after six months suffered a 70% loss, we
might erroneously believe it to be within acceptable limits, while actually,
143
in the first year the drawdowns should probably be a lot less harsh. A case
THE WORK PLAN

such as this should have started an alarm bell ringing previously.
To plan your activities, you should therefore simulate the trend of prof-
its and drawdowns after one year, after two years and so on, for as far as
you intend to go. The simulations will provide information on the potential
results of the strategy, so the effective performance can be monitored during
the period in question.
The thing that’s most worrying is the drawdown that might be suffered
and it’s important to have parameters that can be used to halt trading if
necessary before it’s too late. So you have to set limits on using the strategy,
abandoning it if the real results are within these limits (obviously the
limits for lowest acceptable profits and, above all, the limits for maximum
drawdowns).
If one of the limits set is exceeded, this should ring an alarm bell, as the
results of the strategy are very different to those expected, so it may no
longer be valid. Anyone who knows a bit about trading systems or meth-
ods based on repetitive phenomena knows that sooner or later, things may
change and no longer continue, as they have even for a long time in the past.
At this point, it’s important to pause and take a good look at the whole setup
on which your approach is based. So how do you do this?
By simply looking at the drawdown histograms in the previous chapter,
you’ll notice the distribution in all of them is like that of classic statisti-
cal curves. In particular, the distribution of the percentage drawdowns is
log-normal. For lovers of statistics let’s say the distribution of the natural
logarithm of the percentage drawdowns is normal, so the percentage draw-
downs are distributed log-normally.
This means that in order to calculate an average drawdown, first we have
to calculate the natural logarithms of the various drawdown values, then find
the arithmetic mean, and finally calculate the result of e raised to that mean.
On the basis of the percentage drawdown values DD%1, DD%2, DD%3,
... DD%n we’ll calculate:
In(DD%1) + In(DD%2) + In(DD%3) + ... + In(DD%n)
DDInm =
n
DD%m = eDDInm
where e is equal to 2.7182818, a number maths lovers will be sure to rec-
ognize. These are, in any case, all functions you can find in Microsoft Excel,
144 and the software that comes with the book will run the calculations, to save
the trader who wants to use it the bother.
THE WORK PLAN
Returning to statistical concepts, let’s consider the standard deviation.

This measures the dispersion around a series of mean values of the same.
The greater the standard deviation, the greater the dispersion of the values
around the mean.
A normal distribution has approximately 68% of its values within one
standard deviation from the mean, 95% within approximately two standard
deviations and 99.7% within approximately three standard deviations. (The
concept of the z-score is also based on the standard deviation and the value
1.96 isn’t used by chance. In fact, it guarantees a 95% probability of depen-
dence for the trades.)
In practice, we could calculate the average drawdown and the standard
deviation (also on the basis of the logarithmic series of the drawdowns, to
be carried over for usable values) and estimate zones that are acceptable.
In consideration of the fact that two standard deviations constitute a
boundary within which 95% of the cases considered lie, we could use these
two values as expected limits within which our percentage drawdown
should be after the period in question.
Obviously, every trader is free to calculate the values they want, using
more restrictive limits on the basis of their own particular disposition to risk.
If, for example, the results of the strategy indicated that the average draw-
down in one year of work, should be around 20% with a standard deviation
of 5, this would mean that 95% of the values calculated would be in the
range 20 ± 2*5 or between 10 and 30. If the maximum percentage draw-
down after one year is 34% you might think the strategy is no good and
should be abandoned. To tell the truth, one could say the probability of this
case occurring was less than 5%, and decide whether this is dangerous.
If we considered three standard deviations as an acceptable limit we
would, however, have continued as the upper limit would be 35% (20 +
3*5).
Once you’ve chosen the risk profile and the money management method
you intend to use, it’s a good idea to run simulations for at least one to two
years, taking note of the probable results, especially in terms of drawdowns.
If you decide to abandon a strategy when a drawdown suffered in the field
is more than two standard deviations over the calculated mean, at regular
intervals you would check you’re still within the set limits.
As a concrete example, let’s suppose we’re using a moving average
crossover strategy, and we’re at the end of the third year, so only have the
145
first 480 trades to work with. We’ll plan our trading for the next three
THE WORK PLAN

years on the basis of that data.
So, we’ll run simulations with the 480 trades to choose the most suitable
approach to risk.
Let’s run simulations on 480 trades to see the result after another 480
trades if we used the strategy for three years.
Starting with a hypothesis of using a fixed fractional method based on a
stop-loss with a 5% risk, Figure 7.1 shows the results in terms of profits.
The distribution of the percentage profits is also basically log-normal. The
peaks in the figure between 200% and 250% and between 500% and 1,000%
are due to the fact that, in these zones, the ranges are a lot more extensive,
so naturally they include more data. We could show everything with a 10%
range, but the histogram would be huge and quite difficult to read.
Let’s take a look at what we could expect from the drawdowns.
Note that the value of the drawdowns is basically in the zone around
35%. In particular, the program tells us that the average drawdown is
38.65%, the values of the first standard deviation are shown as between
30.24% and 49.39%, which means that, for approximately 68% of the
THE WORK PLAN
146
0
5
10
15
20
25
30
35
40
45
50
fro
0
10
20
30
40
50
60
70
80
m
1 fro –5 <–5
30
m 0t 0
3 fro –4 o –
m 0 t 40
database.
5 fro –3 o –
11
m 0 t 30
7 –2 o –
16
fro 0 to 20
9 m
FIGURE 7.2
FIGURE 7.1
–
– 1
37
11 fro 10 0
m to
13
41
fro 0 0
m to
15
31
fro 10 10
m to
17
54
fro 20 20
m to
19 fro 30 30 34
m to
21 fro 40 40
50
23 m to
41
fro 50 50
25 m to
33
fro 60 60
27 m to
24
fro 70 70
29 m t
fro 8 o 80
m 0
26
31 fro 9 to
33 m 0 t 90
38
fro 10 o 1
35 m 0 t 00
30
o
fro 11 11
37
23
m 0 to 0
fro 12 12
39
16
m 0 to 0
41 fro 130 13
m
20
to 0
DD distribution
43 fro 14 14
24
m 0t 0
45 fro 150 o 1
m 5
22
to 0
Trend of the drawdowns with a 5% risk.

47 fro 16 16
18
m 0t 0
49 fro 7 1 1 o
21
51 m 0 t 70
fro 18 o 18
53 m 0 to 0
fro 19 19
55 m 0t 0
13 13
fro 20 o 2
57
20
m 0 t 00
59 fro 25 o 2
72
m 0 to 50
61 fro 30 30
m 0t 0
49
63 fro 35 o 3
42
m 0 t 50
65 fro 40 o 4
m 0 00
34
67 fro 4 to 4
m 50 50
69 50 to
0 50
to 0
17 16
71 10
0
Profits after approximately three years with a 5% risk and a three-year
73 > 0
66
10
75 00
16
drawndown results
average drawdown
drawdown +/-1 std dev
hide
FIGURE 7.3 Summary of the drawdown data.
cases, the percentage drawdown values are in this range. With the values of
two standard deviations, we obtain the range of approximately 95% of the
values. This range is from 23.66% to 63.11%.
Figure 7.3 shows a summary.
These values might be acceptable, but let’s see what could happen in these
three years using the fixed ratio method with a delta of 5,000.
147
THE WORK PLAN

120
105
100
93
80
60 57 58
50 49 48
40
40 35
32 32 32
28
26 24
22 21 22 20 23 21
17 19 17
20 15 16 15
10 12 11 11 10
5
0
m 0t 0
m 0 t 40
m 0 t 30
fro 0 to 20
fro 10 0
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 8 o 80
m 0 t 90
m 0 t 00
m 0 to 0
m 0 to 0
to 0
m 0t 0
to 0
m 0t 0
m 0 t 70
m 0 to 0
m 0t 0
m 0 t 00
m 0 to 50
m 0t 0
m 0 t 50
m 0 00
m 50 50
to 0
> 0
00
fro –5 <–5
– 1
fro 11 11
fro 12 12
fro 130 13
fro 14 14
5
fro 16 16
fro 18 o 18
fro 19 19
fro 30 30
0 50
0
fro –4 o –
fro –3 o –
–2 o –
fro 10 o 1
fro 150 o 1
fro 7 1
fro 20 o 2
fro 25 o 2
fro 35 o 3
fro 40 o 4
fro 4 to 4
10
10
m t
m to
m to
m to
m to
m to
m to
m to
fro 9 to
m t
50 to
m 0
m
1
m
m
fro
FIGURE 7.4 Percentage returns for 480 trades using the fixed ratio method with a
delta = 5,000.
DD distribution
50
45
40
35
30
25
20
15
10
0
1
4
7
10
13
16
19
22
25
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
97
28
FIGURE 7.5 Percentage drawdown trend for 480 trades using the fixed ratio method
with a delta = 5,000.
drawndown results
148 average drawdown

THE WORK PLAN
hide
FIGURE 7.6 Data on the distribution of the drawdowns in Figure 7.5.
Here are the same three figures for this case.

Note that in this case, the summary data on the drawdowns is very sim-
ilar for both the fixed fractional and the fixed ratio method, with the latter
slightly more extreme in terms of the limit of the second standard deviation.
The results of the equity, however, make the fixed ratio method the obvious
choice as there is less dispersion in the negative zone and greater solidity in
the zone of limited percentages.
A drawdown of over 60% is still worrying though, so let’s try to run a
simulation with the fixed fractional method and a risk limited to 4%.
Here are the results.
90
79
80
70 67
62
60
60
52 51
50 48 49
44
39 39
40
35 34 35
29 30
30 27
21
20
19 19 18 19 18 20 20 20
14 13
12
10
2 3
0
m 0t 0
m 0 t 40
m 0 t 30
fro 0 to 20
fro 10 0
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 8 o 80
m 0 t 90
m 0 t 00
m 0 to 0
m 0 to 0
to 0
m 0t 0
to 0
m 0t 0
m 0 t 70
m 0 to 0
m 0t 0
m 0 t 00
m 0 to 50
m 0t 0
m 0 t 50
m 0 00
m 50 50
to 0
> 0
00
fro –5 <–5
– 1
fro 11 11
fro 12 12
fro 130 13
fro 14 14
5
fro 16 16
fro 18 o 18
fro 19 19
fro 30 30
0 50
0
fro –4 o –
fro –3 o –
–2 o –
fro 10 o 1
fro 150 o 1
fro 7 1
fro 20 o 2
fro 25 o 2
fro 35 o 3
fro 40 o 4
fro 4 to 4
10
10
m t
m to
m to
m to
m to
m to
m to
m to
fro 9 to
m t
50 to
m 0
m
1
m
m
fro
FIGURE 7.7 Trend of profits with a 4% risk after 480 trades. 149
THE WORK PLAN

It’s immediately obvious that the situation in terms of negative perfor-
mance has improved considerably compared to that in Figure 7.1. In fact,
there are no longer any cases with losses of over 50%. Finally, the number
of cases with modest returns have increased, but there are still a considerable
number of cases with profits up to 100%. Obviously, there are fewer cases
with very high profits, as is to be expected with a lower risk.
Figure 7.8 looks similar to Figure 7.2 but the curve has moved left, which
is what we expected and wanted.
Figure 7.9 is reassuring. If we decide to keep trading only if we remain
within 95% of the expected cases, we’d suffer a maximum percentage draw-
down of 51.69%, which, although notable, is less harsh than the values of
over 60% in the previous simulations.
So we decide to proceed with the fixed fractional method using a 4% risk.
We’ve taken the first step by choosing the risk profile we want to trade on
the market with. The second thing to do is build the plan for growth, step
by step.
DD distribution
60
50
40
30
20
10
0
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
53
55
57
FIGURE 7.8 Trend of the drawdowns with a 4% risk.
drawndown results
average drawdown
150
THE WORK PLAN
hide
FIGURE 7.9 Summary of the data on drawdown distribution with a 4% risk.
Now, let’s see what we might expect after one year (160 trades) and after
two years (320 trades).
We’ll run the simulation again, first with 160 and then 320 trades, then
take a look at the data on equity and drawdowns (Figures 7.10–7.13).
Note that after one year, it’s quite probable that profits will be at least
20% but, above all, we can expect a drawdown of around 18%; we will be
willing to stick it out as long as the maximum drawdown isn’t over 37.16%.
250
229
200
151 146
150
107
100
53
48
50 43 45
35 35
24 21
19
12 13
5 2 4 2 4 1 1
0
m 0t 0
m 0 t 40
m 0 t 30
fro 0 to 20
fro 10 0
fro 0 o 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 8 o 80
m 0 t 90
m 0 t 00
m 0 to 0
m 0 to 0
to 0
m 0t 0
to 0
m 0t 0
m 0 t 70
m 0 to 0
m 0t 0
m 0 t 00
m 0 to 50
m 0t 0
m 0 t 50
m 0 00
m 50 50
to 0
> 0
00
fro –5 <–5
– 1
fro 11 11
fro 12 12
fro 130 13
fro 14 14
5
fro 16 16
fro 18 o 18
fro 19 19
fro 30 30
0 50
0
fro –4 o –
fro –3 o –
–2 o –
fro 10 o 1
fro 150 o 1
fro 7 1
fro 20 o 2
fro 25 o 2
fro 35 o 3
fro 40 o 4
fro 4 to 4
10
10
m t
m to
m to
m to
m to
m to
m to
m to
fro 9 to
m t
50 to
m 0
m
1
m
m
fro
FIGURE 7.10 Expected returns after the first 160 trades (after about 1 year) with a
4% risk.
drawndown results
151
average drawdown
THE WORK PLAN

hide
FIGURE 7.11 Expected drawdowns after the first year with a 4% risk.
After two years, the returns will be a lot more spread out, luckily with
positive results. Obviously, there will also be harsher drawdowns, and in the
second year we should expect to have to suffer a drawdown of up to 46.1%.
We chose this method because we thought we’d be able to withstand
drawdowns of up to about 51%. In the first two years, though, the limit
140
127
124
120
100
84
80
62
60 55 56
50
45 47
42
40
32
28
25 26 25
21 22 21
19
20
13 11
7 9 10 10 9 8 8
1
0
m 0t 0
m 0 t 40
m 0 t 30
fro 0 to 20
fro 10 0
fro 0 0
fro 10 10
fro 20 20
fro 30 30
fro 40 40
fro 50 50
fro 60 60
fro 70 70
fro 8 o 80
m 0 t 90
m 0 t 00
m 0 to 0
m 0 to 0
to 0
m 0t 0
to 0
m 0t 0
m 0 t 70
m 0 to 0
m 0t 0
m 0 t 00
m 0 to 50
m 0t 0
m 0 t 50
m 0 00
m 50 50
to 0
> 0
00
fro –5 <–5
– 1
fro 11 11
fro 12 12
fro 130 13
fro 14 14
5
fro 16 16
fro 18 o 18
fro 19 19
fro 30 30
0 50
0
m to
fro –4 o –
fro –3 o –
–2 o –
fro 10 o 1
fro 150 o 1
fro 7 1
fro 20 o 2
fro 25 o 2
fro 35 o 3
fro 40 o 4
fro 4 to 4
10
10
m to
m to
m to
m to
m to
m to
m to
fro 9 to
m t
50 to
m 0
m
1
m
m
fro
FIGURE 7.12 Expected returns after 2 years of using the strategy with a 4% risk.
drawndown results
152 average drawdown

THE WORK PLAN
hide
FIGURE 7.13 Expected drawdown values after 2 years of using the strategy with a 4%
risk.
is lower and we might only have to withstand this level of drawdowns dur-
ing the third year of using the strategy. If it occurred during the first two
years it would be better to cease trading.
Now let’s see what happened after the first 480 trades when the plan had
been prepared, in a period of actual trading.
0,00%
100.000,00
5,00%
90.000,00
10,00%
15,00% 80.000,00
drawdown %
20,00%
equity
70.000,00
25,00%
60.000,00
30,00%
50.000,00 50.000,00
35,00%
40,00% 40.000,00
FIGURE 7.14 Trend of profits and drawdowns during the first year.
We’ll proceed year by year and first we’ll take a look at the situation after
placing the first 160 trades.
Figure 7.14 shows the trends of the drawdowns and profits superposed.
153
Note that the year closes just over €90,000 at €92,300. The maximum per-
THE WORK PLAN

centage drawdown of 13,46% was in the sixtieth trade. Note that this isn’t
the maximum drawdown in terms of absolute value. Also, at a glance, in
fact you can see that harshest drawdown from the maximum levels reached
is in the second part of the year – but we chose (for the reasons explained
in detail in previous chapters) to use the percentage drawdown as a measure
of the level of sufferance.
According to the values in Figure 7.11, we could ‘suffer’ a drawdown
of up to 37.16% so we are well within the limits. Actually, these values
look particularly benevolent, below the average and near the first standard
deviation.
The particularly favourable period is also confirmed by the results of
the percentage profits, with profits just over 80% and as can be seen in
Figure 7.10 very few cases above these values (a total of 65 starting with 21
in the 80% to 90% range).
Let’s continue to incorporate also the second year of results.
Figure 7.15 shows one of those bad periods that every trader finds himself
in sooner or later. After having reached a peak of around €150,000 at the
0,00%
180.000,00
10,00% 170.000,00
160.000,00
20,00%
150.000,00
140.000,00
30,00%
130.000,00
40,00% 120.000,00
drawdown %
110.000,00 equity
50,00%
100.000,00
60,00% 90.000,00
80.000,00
70,00%
70.000,00
60.000,00
80,00%
50.000,00 50.000,00
90,00% 40.000,00
FIGURE 7.15 Trend of the equity and drawdowns after the second year of actual
trading.
beginning of the second year, we suffered a period of losses that reduced the
equity to less than €85,000, which is less than what we started the year with.
154 A period like this would put the nerves of any trader to the test and many
would probably throw in the towel before they reached the lowest point.
THE WORK PLAN
So why did we continue? It’s simple. In Figure 7.13, we had an

‘acceptable’ drawdown limit of 46.10%. Well, the maximum drawdown in
the worst period for the system reached 45.9%! (This result came really
close to the limit merely by chance, and there was no intention of using
these values to astonish the reader.) We were going to call it a day. Another
losing trade and we probably would have passed the barrier we’d set, and at
these levels we certainly wouldn’t have made any excuses and extended the
limits of tolerance to three standard deviations, so we would have closed up
shop; but at 45.9%, we decided it was still acceptable to continue, and the
plan that had been prepared proved to be right with a subsequent recovery
that closed the year at €102,875 with profits of just over €10,000. Of
course, €10,000 isn’t much compared to the €42,500 of the previous year
(we started with €50,000), but it did make the close of the year a positive
one, and kept hopes up for the future.
Now let’s see the outcome after the three forecasted years (Figure 7.16).
0,00%
240,000,00
230,000,00
10,00% 220,000,00
210,000,00
20,00% 200,000,00
190,000,00
180,000,00
30,00%
170,000,00
160,000,00
40,00%
150,000,00
drawdown %
140,000,00 equity
50,00% 130,000,00
120,000,00
60,00% 110,000,00
100,000,00
90,000,00
70,00%
80,000,00
70,000,00
80,00% 60,000,00
50.000,00 50,000,00
90,00% 40,000,00
FIGURE 7.16 Final equity and drawdowns after three years of real trading.
At the end of our adventure (we decided to use the strategy for three
years) we have an equity of €165,015.5, with profits of around 230% on the
initial capital of €50,000. The maximum drawdown is the above-mentioned
45.9%. Note that at the beginning of the third year the equity drops again,
but this time it stops at €95,000 to then recover and finally let us breathe easy. 155
The 230% profits are at the limit of the first standard deviation of the
THE WORK PLAN

values in Figure 7.7 (this figure is calculated by the software, which gives
an extremely wide range of 54,107 – 155,604 for the first standard devia-
tion) and is therefore in line with the best estimates. As for the drawdown
suffered, 45.9% was particularly harsh in the first two years, but now this
can be considered in perspective, as it’s lower than the 51.69% estimated as
being the maximum allowed drawdown (Figure 7.9).
A well-constructed work plan helped bring our work to a successful con-
clusion and, above all, get through a very ugly period without getting into a
panic (there were of course a few frayed nerves, but having a plan helped in
the decision-making process).
■ 7.2 Conclusions
The correct use of the tools at our disposal to simulate how a strategy might
behave can help in the decision-making process when considering whether
to continue or cease trading with the same strategy. Meticulously preparing
a plan that clearly sets limits of acceptance and those that need monitoring,
before starting, helps the trader at difficult times when you might fall prey
to your emotions.
At this point, you must, of course, be unmoving in your resolve and stick
to the set guidelines, because sometimes you can be walking the razor’s edge,
and only sticking faithfully to the rules will let you continue and conclude
the work you started with the best possible outcome.
With the Monte Carlo simulation, you can prepare possible scenarios
after various periods of time using the strategy, and can prepare a list
of potential profits (in general, these are much more difficult to predict
when using more aggressive strategies), and potential drawdowns (to be
considered on the basis of how difficult it will be for the trader to withstand
the same).
156
THE WORK PLAN
CHAPTER 8
Combining Forces
At this point you could stop reading and start working, but first let’s take a
detailed look at some of the other things you might encounter as a trader.
One theme that’s definitely worth looking into is diversification, how
combining various strategies improves the overall situation.
We’ve seen that the aim of applying money management techniques is to
find the best amounts to adopt to maximize profits while limiting drawdowns
to acceptable values. 157
■ 8.1 Using a Combination of Systems

Up to now we’ve worked with a strategy that, while it certainly can’t be
considered excellent, is quite suitable for analyzing most of the techniques
we’re interested in. Now let’s take a look at what would happen to the equity
lines if we combined various systems.
To do so we’ll use a new system, which this time is based on hourly bars,
and simply buys on the breakout of the 50-bar high and sells on the breakout
of the low in the same period.
This concept, known as Donchian Channel Breakout, is one of the most
classic trend-following models found in literature, and it’s based on the con-
viction that when a market is strong enough to exceed the highest highs of a
certain period, then it is highly probable that it will continue in that direc-
tion (and therefore give us the chance to follow a long trend) and vice versa,
on a breakout of the lowest lows in the same period (50 bars, in our case)
it shows signs of weakness that suggests further down movements will follow
(and it’s therefore a good opportunity for trading short).
We’ll apply the model to three futures’ markets that usually have good
trending characteristics, the Crude Oil, Gold, and Soybean markets.
The strategy will be applied with a 1,500 USD stop-loss per contract to
protect the position.
First, let’s look at results on the crude oil market only, in the period from
the 1 January 2010 to the end of 2017 (Figures 8.1–8.3).
The trend of the strategy isn’t exactly exceptional despite for the fact that
there are profits in time. If we look at the results year by year, we’ll see that
some years close with a loss.
Now let’s take a look at the results using the same approach with Gold
futures (Figures 8.4–8.6).
As can be seen, the results in numerical terms for Gold and Crude Oil
are very similar, although the trend of gold is much more regular over the
years.
158
COMBINING FORCES
FIGURE 8.1 Crude Oil strategy performance report.

Equity Curve Detailed with Drawdown
180,000
160,000
Equity ($)
140,000
120,000
100,000
Drawdown ($)
–20,000
–40,000
Drawdown (%)
–10
–20
–30
9/13/2011 21:00:00 5/24/2013 06:00:00 2/3/2015 08:00:00 10/12/2016 05:00:00
Date
FIGURE 8.2 Crude Oil equity line.
159
COMBINING FORCES
FIGURE 8.3 Annual result for Crude Oil
As mentioned above we’ll also consider Soybean futures and verify the
system metrics on that market (Figures 8.7–8.9).
In this case, there’s a flattening in the returns in the later years after a
start that produced higher profits (but also a harsher drawdown in the initial
period).
FIGURE 8.4 Gold future strategy performance report.
160
COMBINING FORCES
175,000
Equity ($)
150,000
125,000
100,000
Drawdown ($)
–10,000
–20,000
Drawdown (%)
–5
–10
–15
9/13/2011 20:00:00 5/24/2013 04:00:00 2/3/2015 07:00:00 10/12/2016 04:00:00

Date
FIGURE 8.5 Gold equity line.

FIGURE 8.6 Annual result for Gold.
161
COMBINING FORCES
FIGURE 8.7 Soybean strategy performance report.

160,000
Equity ($)
140,000
120,000
100,000
Drawdown ($)
–10,000
–20,000
Drawdown (%)
–10
2/14/2012 02:00:00 2/25/2014 03:00:00 4/4/2016 20:00:00

Date
FIGURE 8.8 Soybean equity line.
162
COMBINING FORCES
FIGURE 8.9 Annual result for Soybean.
For the first analysis on the concept of collaboration between systems,

we’ll combine the three systems in a portfolio in the hypothesis of invest-
ing with an initial capital of 100,000 USD to check the resulting behaviour
(Figures 8.10–8.12).
FIGURE 8.10 Portfolio of systems without money management.
300,000
250,000
Equity ($)
200,000
150,000
163
COMBINING FORCES
Drawdown ($)
–20,000
–40,000
Drawdown (%)
–10
–20
9/14/2011 01:00:00 5/24/2013 09:00:00 2/3/2015 11:00:00 10/12/2016 08:00:00

Date
FIGURE 8.11 Equity line of the portfolio of systems without money management.
The turbulence in the initial years has an effect also on the performance
of the portfolio, but in the last periods there’s a certain improvement in the
trend of the equity line with evident greater stability in terms of drawdowns.
The percentage drawdown figure doesn’t make much sense, as it’s calculated
each time on the basis of the level the equity line has reached at that partic-
ular time, so it depends on accumulated profits and when the analysis starts.
FIGURE 8.12 Annual trend of the portfolio of systems without money management.
The absolute drawdown does, however, provide a valid point of reference

and is a measure of the stability already mentioned. We can make other
considerations on the lines we’ll draw when we analyze the application of
164 the money management models.
Now, to evaluate the effect of both the application of position sizing on
COMBINING FORCES
the systems, and whether a portfolio approach is effective or not, first we’ll
consider the individual systems applying different money management algo-
rithms, starting with the Crude Oil system and evaluating the performance
with a percent f of 2% calibrated to the stop-loss, a fixed ratio with a delta
of 10,000 USD and a percent volatility of 2% (Figures 8.13–8.15).
No matter how you look at it, you’ll see that the Crude Oil performance
shows no substantial improvement after applying money management. On
the contrary, with some models the performance is even worse than the
single contract model.
The cause of this is the irregularity of the equity line and the volatility
of the instrument, evidently too aggressive for a capital of 100,000 USD
(partial confirmation of this comes from the fact that the percent volatility
model is the only one for which performance has improved with one single
contract and, as we know, this model prevents trading in market conditions
that are too turbulent for the available capital).
FIGURE 8.13 Performance report for the system with Crude Oil and a percent f
of 2%.
165
COMBINING FORCES
FIGURE 8.14 Performance report for the system with Crude Oil using the fixed ratio
method with a delta = 10,000 USD.
FIGURE 8.15 Performance Report for the system with Crude Oil and a percent
volatility of 2%.
In any case, let’s take a look at the results, applying the same position
sizing models to the Gold futures market (Figures 8.16–8.18).
The situation isn’t exactly idyllic for the Gold futures, either. Using the
fixed ratio method there are no results (Figure 8.17), and only the percent
volatility method shows some improvement, although in this case probably
due to the halt imposed when the market is too turbulent.
Now let’s take a look at Soybean futures, applying the same position sizing
algorithms (Figures 8.19–8.21).
166
COMBINING FORCES
FIGURE 8.16 Performance report for the system with Gold futures and a percent f of
2%.
FIGURE 8.17 Performance report for the system with Gold futures using the fixed
ratio method with a delta = 10,000 USD.
FIGURE 8.18 Performance report for the system with Gold futures and a percent
volatility of 2%.
167
COMBINING FORCES
FIGURE 8.19 Performance report for the system with Soybean futures and a percent f
of 2%.
FIGURE 8.20 Performance report for the system with Soybean futures using the fixed
ratio method with a delta = 10,000 USD.
FIGURE 8.21 Performance report for the system with Soybean futures and a percent
volatility of 2%.
‘Finally,’ one of the systems shows signs of improvement on all fronts,

although still nothing to write home about; but in comparison to the disap-
pointing Crude Oil and Gold results the Soybean results prove the effective-
ness of money management.
■ 8.2 Portfolio Money Management

168
We’ve reached the point where it’s time to put everything we’ve learnt into
COMBINING FORCES
practice and consider using a portfolio.

So what’s the best way to proceed now we have three instruments? How
does the approach change?
Substantially, there’s little change to the logic illustrated in the previous
chapters. We’re still thinking in terms of risk percentages for the fixed frac-
tional method and in terms of delta for the fixed ratio method, while for
percent volatility we’ll consider the usual allowable percentage fluctuation.
Also in this case we’ll use the system stop-loss or the largest losing trade
actually recorded in the period being studied as the maximum loss. In both
the first and the second case, it’s important to use the values that refer to
each system/instrument on the basis of the input signal received. In our
case, using the stop-loss eliminates any problems as the value is the same for
all instruments.
Someone might object that we’re using the profits from one instrument
to increase the number of contracts on another, or that the performance of a
strong system is hampered by the losses of another system in crisis with the
number of its contracts increased thanks to the first.
In money management, you don’t think in terms of single systems. Always
remember that the number of contracts depends on the available capital. If
you’re using the fixed fractional method and have 100,000 USD and use one
contract, you’ll use two (or even three, depending on how much that one
for the first contract was rounded down) if you have 200,000 USD. So if
you start with 100,000 USD, and the system performs well from the start
and brings your equity to 200,000, it’s only natural for the second system
to take that 200,000 as a reference, just as if you’re starting from that point
with that capital.
The fixed ratio method is slightly different. In this case, when you start
using the system you’ll start with just one contract, so the profits accumu-
lated using other instruments shouldn’t be taken into consideration when
calculating the number of contracts to use in the first trade. In actual fact,
starting with just one contract is something you can of course decide to
change. The basic concept is that, without prejudice to the initial capital,
you increase the number of contracts in a way that’s more or less linked to
what you make on the market. In fact, you’d go back to using one contract if
your equity dropped back to below the level of your initial capital. Follow-
ing this line of reasoning, starting with 100,000 USD, once we’ve reached
200,000 USD perhaps thanks to the Soybean system, we’d simply calculate
the number of contracts to use with the Gold system as if it produced those 169
COMBINING FORCES
profits. In fact, also the Gold system in practice starts with 100,000 USD,
and we’d adjust the number of contracts from this value on, so at 200,000
we’d run the necessary calculations and wouldn’t consider this as the initial
capital any longer.
■ 8.3 Which Capital?

This matter is more complex than one might think. Which capital should be
considered when calculating the number of contracts?
When we were working with just one single instrument and a single sys-
tem, the situation was simple enough. You calculate the capital at the close
of the trade and use it as the point of reference for calculating the number of
contracts for the next trade. Cases with reverse orders might seem complex,
but using the correct setup for the calculations lets you bypass this obstacle.
Now we’re using several instruments at the same time, and one of these
might be in position when there’s a signal to enter a trade using another
instrument.
Various methods can be used as ways to consider your capital of reference,
and everyone is free to choose the one they find most suitable.
The first method, invented by Van Tharp called the core equity model,
considers the available capital to be that remaining after deducting the poten-
tial losses allowed by the money management strategy.
In practice, if, for example, you have an available capital of 500,000 USD
and adopt the above strategy with Gold using a percent f of 2%, you’d be
able to enter the trade with six contracts:
capital risked = 500,000 ∗ 2∕100 = 10,000
contracts = ENT(10, 000∕1,500) = 6.
This means you’d be willing to withstand a loss of 9,000 USD. Now, if
you received a signal to enter a Crude Oil trade, you’d consider your capital
to be the initial 500,000 USD less the 9,000 risked and would therefore use
the value 491,000 to calculate the number of Crude Oil contracts.
A second Van Tharp method is called the total equity model. In this case,
the capital of reference is the total available liquid capital added to the cur-
rent value of the portfolio. Taking the same previous example, let’s suppose
you have a gain of 2,000 USD for each Gold contract in position when you
170 decide to enter the Crude Oil trade. In this case the capital to be considered
would be
COMBINING FORCES
500,000 + 6 ∗ 2,000 = 512,000.

It’s immediately evident that the situation is quite different, and a different
choice could produce very different results.
Van Tharp also came up with a third method called the reduced core
equity model, although he later changed the name to reduced total equity
model. This method is based on the reasoning behind the core equity model
and subtracts the risk values from the initial capital. If some values change
while the position is open, however, as a result of a dynamic trailing stop or
take profit, the method returns the staked parts to the value of the capital to
be considered.
If, for example, in the above case you reached a point in which the Gold
contract provided for a move to a closer exit level (this isn’t provided for
by the system we’re taking as an example but is often used in many other
systems), and instead of a loss of 1,500 USD now provided for a loss of
500 USD, the 1,000 USD ‘gained’ from the dynamics of the trade for each
contract would be added to the value of the capital to be considered.
So before we had:
500,000 − 6 ∗ 1,500 = 491,000.
Now we have:
491,000 + 6 ∗ 1,000 = 497,000.
We’re simply subtracting what remains as the risk of a loss from the total
capital, in other words 500 USD for each contract trading, in fact:
500,000 − 6 ∗ 500 = 497,000.
The substantial difference with this approach, though, is when another
increase in profits from the Gold contracts once again results in changing
the exit level, to this time guarantee: for example, 500 USD of profits from
each contract.
At this point, we’ll have changed the level of risk by another 1,000 USD
per contract (from a 500 loss to 500 in profit) and based on the previous
497,000 of reference, we’d calculate:
497,000 + 6 ∗ 1,000 = 503,000,
which also corresponds to the sum of the capital and the profits guaranteed
171
by the open trade, which, we’ve assumed, is equal to 500 USD per contract.
COMBINING FORCES
500,000 + 6 ∗ 500 = 503,000.
The last method (the last we’ll look at in this book, but certainly not
the last in terms of possibilities, which are limited only by the trader’s
imagination) is, for every trade, to consider the equity when the last trade
closed. In practice the evolution of the open trades and their potential
repercussions on the capital, aren’t considered. As far as I know, this
method doesn’t have a name, and I do sometimes use it as the calculations
are intrinsically very simple.
Apart from the implications in terms of results, there are also practical
aspects to consider when deciding whether to use one method rather than
another.
The core equity model is quite well-suited to evaluating series of trades in
Excel. If you have the date and time when each trade was opened and closed,
it’s easy to build the above-mentioned model. You can’t go further without
the detailed history of the evolution of the trade, so you can’t apply the total
equity model or the reduced total equity model.
The total equity model, however, is easy to use for those who can monitor
their portfolio in real time, and therefore have a constantly updated value
of the amount of cash and the projection of the positions. Not all trading
platforms make this information available for it to be used immediately, so
each individual should see whether they can use the model described, and
above all, what sort of ‘feeling’ they have for it.
The reduced total equity model is definitely more complex but lets you
follow the trend of your system in the best possible way. While the core
equity model considers the limits of the trade (stop-loss or maximum loss)
only at the beginning of the trade, for the reduced total equity model you
have to record the evolution of said parameters during the entire life of the
trade. If a stop is increased in the base system, the model must also take this
into account.
The last method described comes in handy if you keep accounts of closed
trades, which are only updated when each trade does in fact close. This,
therefore, refers to the total equity, which is updated each time a new
amount is added or subtracted.
It can be said that the core equity model is certainly the most conservative.
In fact, every time you open a position it considers the worst event. In other
words, it considers your capital to be that which remains after docking the
172
estimated loss for that position. As we’ve seen, a more conservative approach
COMBINING FORCES
can often offer some protection from adverse events, but it’s also important
to consider that this approach can limit the potential of the system in the
case of long-term trades.
In fact, let’s suppose you open a position for a month. Obviously, during
that month the position could be very profitable, and using the core equity
model for the whole month it’s as if you have the initial capital after docking
the estimated loss (let’s suppose 1,500 USD per contract) available for the
open position. If you opened a second position after 20 days, the number of
contracts of this position would therefore refer to the portion of your capital
earmarked for the loss of the first position. Let’s imagine, though, that the
first position is doing extremely well and produces open profits of 10,000
USD per contract. You use the core equity model to calculate the number
of contracts for the second position:
500,000 − 6 ∗ 1,500 = 491,000.
With the total equity model, on the other hand, this would give you:
500,000 + 6 ∗ 10,000 = 560,000.
As is easy to imagine, it’s highly probable you’d end up with quite a dif-
ferent number of contracts.
Personally, I also use another method to assess the impact of money
management on portfolios of strategies, which consists of calculating the
total equity at the close of the day before any new trades are entered. This
approach is an approximation (from the previous close to the moment
when a new trade is placed the available capital will certainly change), but
it makes it easy to manage placing contracts as they are fixed at the start of
each day, and there’s no risk of having to calculate them at the same time as
entering the trade with the same.
■ 8.4 The Effects of Portfolio Money

Management
So, now let’s take a look at what effect applying known money management
strategies can have with our system of three instruments.
We’ll apply the same models to each single instrument and calculate the
total equity as described at the end of the previous paragraph.
We’ll start using percent f with a 2% risk and the loss based on a stop-loss
of 1,500 USD for all instruments (Figure 8.22). 173
It’s immediately obviously that the net profit is much higher than the sum
COMBINING FORCES
of the three single results. It’s over 530,000 USD, while the sum of the three
single systems, adopting the position sizing model we’ve used here on each,
is approximately 250,000 USD; we’ve more than doubled the final result!
FIGURE 8.22 Performance report of the portfolio with a percent f of 2%.

The DD has also been affected, and from just over 30%; in this case, it’s
increased to over 40%.
Let’s take a look at the trend of the equity line (Figure 8.23).
It’s an interesting trend. The maximum percentage drawdown (remem-
ber that, in the evaluation, with position sizing the DD% is more representa-
tive than the absolute drawdown, unlike when considering a single contract)
is in the initial phases, where there were difficulties, especially with the
Crude Oil system – while it stabilizes at around 20% in the subsequent
phases.
A look at the results year by year shows good progression, with again the
false note of 2015 that closes at a loss.
Now, let’s take a look at the results when applying the fixed ratio method,
again with a delta of 10,000 USD. (This was chosen as it represents a value
that’s on average near the required margins of these instruments and there-
fore in line with the consequent trading.)
These results are to a certain extent surprising, as the fixed ratio method
wasn’t exactly the highest performer when applied to individual systems;
but, when applied to a portfolio, it produces a notably higher final profit
compared to the same portfolio with percent f. The drawdown is worse,
174
COMBINING FORCES
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–50,000
–100,000
Drawdown (%)
–20
9/14/2011 01:00:00 5/24/2013 09:00:00 2/3/2015 11:00:00 10/12/2016 08:00:00

Date
FIGURE 8.23 Equity line of the portfolio with a percent f of 2%.

FIGURE 8.24 Annual performance of the portfolio with a percent f of 2%.
175
COMBINING FORCES
FIGURE 8.25 Performance report of the portfolio using the fixed ratio method with a
delta of 10,000 USD.
however, and over 50%, which would certainly be difficult for even the most
tenacious of traders to withstand.
Figure 8.26 shows the trend of the equity line and Figure 8.27 shows the
profits year by year.
As can be seen in Figure 8.26, there were prolonged periods of significant
drawdowns and, in the last period, with levels over 25% while percent f
showed levels around 20% only.
Figure 8.27 shows that 2012 would have also closed with a loss.
800,000
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–100,000
–200,000
Drawdown (%)
–25
–50
9/14/2011 01:00:00 5/24/2013 09:00:00 2/3/2015 11:00:00 10/12/2016 08:00:00

Date
FIGURE 8.26 Equity line of the portfolio using the fixed ratio method with a delta of
10,000 USD.
176
COMBINING FORCES
FIGURE 8.27 Annual performance of the portfolio using the fixed ratio method with a
delta of 10,000 USD.
FIGURE 8.28 Performance report of the portfolio with a percent volatility of 2%.
These evaluations weaken the case for using the fixed ratio method,
despite the better performance in terms of final returns.
But let’s continue and take a look at the results using a 2% percent volatil-
ity model (Figure 8.28).
Well, the model that attempts to weigh entering trades to avoid suffering
overly from movements in more turbulent times produces results that are
astonishing, to say the least, when compared to the other approaches.
Now, let’s take a look at the trend of the equity line and the annual distri- 177
bution of profits (Figures 8.29, 8.30).
COMBINING FORCES
1,500,000
Equity ($)
1,000,000
500,000
Drawdown ($)
–200,000
Drawdown (%)
–10
–20
–30
9/14/2011 01:00:00 5/24/2013 09:00:00 2/3/2015 11:00:00 10/12/2016 08:00:00

Date
FIGURE 8.29 Equity line of the portfolio with a percent volatility of 2%.
FIGURE 8.30 Annual performance of the portfolio with a percent volatility of 2%.
It’s immediately obvious that the drawdown in the initial years is more
contained, and finally there are no years that close with a loss (although 2015
can practically be considered to have broken even). It’s interesting to see
178 that, in Figure 8.28, the maximum drawdown is just over 30%.
Having judged the fixed ratio method too ‘unpredictable’ in terms of
COMBINING FORCES
drawdowns, we could compare the percent f of 2% with the percent volatil-

ity method that produces similar results in terms of net profit to compare
the risks for the same results.
In order to produce similar results, we can trade by limiting the fluctu-
ations to less than 1.55%. This choice (instead of 2%) produces the results
shown in Figures 8.31–8.33.
FIGURE 8.31 Performance report of the portfolio with a percent volatility of 1.55%.
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–50,000
–100,000
Drawdown (%)
–10
–20
9/14/2011 01:00:00 5/24/2013 09:00:00 2/3/2015 11:00:00 10/12/2016 08:00:00

Date
FIGURE 8.32 Equity line of the portfolio with a percent volatility of 1.55%.
179
COMBINING FORCES
FIGURE 8.33 Annual performance of the portfolio with a percent volatility of 1.55%.
The most interesting piece of information is provided by Figure 8.31,

where we can see that, with results for profits close to those using a percent
f of 2%, the drawdown is less than 25%!
A result of this kind tips the scales in favour of the percent volatility
model.
■ 8.5 Conclusions
The aim of this chapter was, first and foremost, to show you how combin-
ing several strategies that refer to the same capital increases the potential
of money management methods exponentially. In fact, the number of con-
tracts are gradually increased in a more marked way the more trades there
are. (Paradoxically, imagine a strategy that makes just one good trade: there
would be no point in applying a money management method to this strategy,
as there are no other trades for which you could use different numbers of
contracts.) Combining several strategies increases the total number of trades
and, therefore, intensifies the effect described.
We looked at how various methods might behave on moving average
crossover strategies and drew our conclusions. These conclusions should
not be taken as gospel for all types of strategies, though. What matters is
the process used to reach this point, and it’s this process that must be learnt
in order to be able to replicate it with other systems.
■ Reference
180
2007.
COMBINING FORCES
CHAPTER 9
Money
Management
When Trading
Stocks
181
Until now, we’ve focused on trading futures. These instruments are often
erroneously considered risky, but the real risk is not knowing them, and
therefore not being able to assess the risks. If you know the risks, and take
steps to keep them under control, derivatives are nothing more nor less than
simple trading instruments.
In any case, what we’ve discussed up to this point also applies to other
trading instruments, and in particular to the stock market, which remains
as always the most popular and populated trading market in the world.
■ 9.1 Trading in the Stock Market

When choosing a market to trade on a trader should consider all the pros
and cons, and know what sort of limits they might encounter. In the case of
the futures’ market, often there are limits in terms of the margins required
to trade with a certain instrument, and if the available capital isn’t enough
to cover these margins trading will be blocked. On the stock market we
could say this limit is all but nonexistent: you can buy as many shares as you
want and therefore size the trade you’re entering in a way that doesn’t create
obstacles (unless, of course, the price of the share chosen is already higher
than the capital in your trading account).
In many cases, you can also trade using leverage; in other words, buy (or
sell) more stocks than you could with the capital in your account. There are,
of course, a series of limits to be observed, but this option can certainly offer
new levels of freedom that could be useful.
Trading on the stock market in fact involves an approach that’s instinc-
tively different to trading derivatives. In fact, when buying stocks it’s natural
to think in terms of exposure, so the question you need to answer is, ‘How’
much money should I ‘invest?’ With derivatives, one’s reasoning is almost
always along other lines, and if you’re buying a DAX contract it would be
unlikely for you to think how much it’s really worth in monetary terms.
In other words, you won’t calculate the result of multiplying the current
price for the BigPointValue (value of each point), not because it’s wrong
but because it simply doesn’t come spontaneously; but, it’s exactly this lack
of spontaneity that makes it almost always easier to think in terms of how
182
much you could lose on each trade you’re going to open. With derivatives
MONEY MANAGEMENT WHEN TRADING STOCKS
in general, we think of where to enter and exit trades and how much the
difference between the two levels costs for each contract purchased. So, we
start on the basis of a concept of potential loss and arrive, as a consequence,
at the point in which we can size the position. (In reality, I say this in the
hope that this is what the reader who has reached this point is now doing,
because I know many beginners use much more hazardous approaches. But
in this case, let’s assume one is trading on the basis of the notions acquired
in the previous chapters of the book.)
On the stock market, you’d usually think of how much of your available
capital you’re going to invest and, while this may be considered the correct
approach (if you answer the question allowing for the risk), it is still quite
hazardous.
On the basis of this premise, let’s make a hypothesis that we have just
one system and want to trade any stock on the US market – Amazon, for
example. Well aware of the mean-reverting nature of the principal stocks
on the US market, we decide to trade after a significant fall in the market,
counting on a rebound.
To develop the trade we use intraday data and refer to the daily time frame
to construct our setup.
If, at the close of the previous day, the price has dropped to a new low
over the last three days, we decide, today, to buy at yesterday’s low price, but
we don’t send the entry order right away, we keep it active only from 10:00
to 14:30 (remember the US stock exchange trades from 9:30 to 16:00) to
avoid entering the trade at the time of maximum market volatility.
The position will be closed, either by a 2% stop-loss or a take profit that’s
double the stop.
Figure 9.1 shows an example of the trade.
183

FIGURE 9.1 Entry at lowest low of the previous three days and take-profit exit.
The following EasyLanguage code corresponds to the above concepts.
if low of data2 <= lowest(l of data2,3) and time > 1000 and time
< 1430 then buy next bar at (low of data2) limit;
if marketposition > 0 then sell("LXSL") next bar at entryprice
*(1-Mypercloss/100) stop;
if marketposition > 0 then sell("LXTP") next bar at entryprice
*(1+(Mypercloss*2)/100) limit;
We start trading in the hypothesis of having an initial capital of 10,000
USD paying 10 USD per contract in commission. The first assessment will be
done imagining we use all the available capital; this choice, which might seem
absurd, isn’t exactly ill-advised because we’ve decided to use a 2% stop-loss;
so, trading with all the available capital is the equivalent of an own risk of
2% per trade which, as we’ve seen in the previous chapters, is absolutely
acceptable.
Figure 9.2 shows the performance report, the subsequent trend of the
equity line and returns year by year. Results are appreciable.
Some may feel a drawdown that’s often over 20% is too much, and there-
fore prefer a more conservative trading strategy.
Above, I said that always using all the available capital is the equivalent of
a 2% risk per trade. If we want to use a much less aggressive strategy, we
could risk just 1%; in other words, use only half the capital available to open
positions. What would happen if we halved our exposure? Let’s take a look
at the performance report in Figure 9.5.
Perhaps many will be astonished to see no results when, with a bigger risk,
we obtained such a satisfactory performance. In the previous chapters, we
184
FIGURE 9.2 Performance report starting with 10,000 USD always investing all the
capital.
20,000
Equity ($)
15,000
10,000
Drawdown ($)
-2,500
-5,000
Drawdown (%)
-10
-20
5/17/2006 13:20:00 12/8/2008 15:10:00 6/28/2011 13:55:00 1/22/2014 11:15:00 8/10/2016 13:00:00
Date
FIGURE 9.3 Equity line starting with 10,000 USD always investing all the capital.
185
FIGURE 9.4 Annual profits starting with 10,000 USD always investing all the capital.
saw that, by pressing harder on the accelerator, we could earn more but also
suffer more marked drawdowns; but we never came across a case in which
a greater risk with a losing system could suddenly start making profits. So
how can we explain this phenomenon?
Comparing Figures 9.5 and 9.2 you’ll see they have one thing in common,
and that is the total commission paid.
When we were trading futures we paid a certain amount per contract,
and increasing or decreasing the exposure as a consequence increased or
decreased the commission paid; on the stock market on the other hand, apart
from some exceptions that can in any case be even more costly, a per-trade
commission is paid. It doesn’t matter whether said trade is for a lot of 10,000
USD or 5,000 USD, and this is what makes a difference using our approach:
the impact of the 10 USD commission, set to simulate realistic calculations,
becomes too burdensome when opening smaller positions. In short, a single
trade can often be invalidated, or even make a loss, due to the impact of the
commissions, and this can destroy what would otherwise be a good system.
I don’t want to imply that trading derivatives is cheaper and trading stocks
is too costly, but I do want to emphasise you should always consider the
186
FIGURE 9.5 Performance report with a 1% risk per trade (exposure 50% of available
capital).
various scenarios and never underestimate anything that could in any way
throw a spanner in the works.
I mentioned other commissions, and some brokers offer proportional
transaction costs (in other words, the commission is calculated in proportion
to the sum traded) but apart from some very rare exceptions (the only bro-
ker that comes to mind is the Italian broker Directa, who may have changed
their conditions in the meantime) these commissions do have a minimum
fee: which, again, has a negative effect when trading smaller amounts.
At this point, we could obviously consider doing the opposite and increase
the risk to minimize the impact of commission. While, in mathematical
terms, this idea holds water, it does, however, go against risk containment
logic, and there might also be problems with the limits imposed on trading
by the broker you’ve chosen.
In fact, if the broker doesn’t allow any kind of leverage, it will be impos-
sible to open positions that could result in exposure to risks of over 2%, as
this is how much we’d lose using all the available capital (as it coincides with
the percentage stop of the strategy). If we wanted to risk 3% per trade we’d
have to open positions that, if stopped at 2% from the entry point, would
lose 3%, so positions 1.5 times the whole account, and this could only be 187
done with leverage.

Let’s imagine, just to see what might happen, that we could use 200%
leverage and therefore could adopt a 3% risk per trade. Figure 9.6 and the
following shows the results we would have obtained.
These are very interesting results, although the drawdown is definitely
starting to be too much. It would be interesting to see what sort of impact a
percent volatility model might have, and whether it was more or less effec-
tive. To obtain a performance near that with a percent f of 3% in Figure 9.6,
we’d have to use a volatility percentage of 4%, which produces the results
shown in Figure 9.9 and the following.
Note that the max. drawdown with percent volatility is higher than that
with percent f for the same results. Comparing Figures 9.7 and 9.10, you’ll
see that, with percent f the drawdown is more regular, while with percent
volatility there are two spikes and then lower values on average (around
20%). It’s down to the individual trader to decide whether they prefer one
approach or the other. Leverage for both was 200%.
FIGURE 9.6 Performance report with a 3% risk per trade (exposure 150% of
available capital).
188
50,000
40,000
Equity ($)
30,000
20,000
10,000
Drawdown ($)
-5,000
-10,000
-15,000
Drawdown (%)
-20
5/17/2006 13:20:00 12/8/2008 15:10:00 6/28/2011 13:55:00 1/22/2014 11:15:00 8/10/2016 13:00:00
Date
FIGURE 9.7 Equity line with a 3% risk per trade (exposure 150% of available capital).
189
FIGURE 9.8 Annual profits with a 3% risk per trade (exposure 150% of available

capital).
FIGURE 9.9 Performance report with a percent volatility of 4%.

50,000
Equity ($)
40,000
30,000
20,000
10,000
Drawdown ($)
-5,000
-10,000
-15,000
Drawdown (%)
-20
-40
5/17/2006 13:20:00 12/8/2008 15:10:00 6/28/2011 13:55:00 1/22/2014 11:15:00 8/10/2016 13:00:00
Date
FIGURE 9.10 Equity line with a percent volatility of 4%.
190
FIGURE 9.11 Annual profits with a percent volatility of 4%.

Here’s the complete code, with all the variables and inputs you can use
for various evaluations.
var: myshares(1000), MyATR(1);

var: MP(0), MyEquity(10000);
MP = marketposition;
input:MyInitialBalance(10000),Mypercloss(2), Myrisk(1), MM(1),
percvol(2), marg(100);
MyATR= AvgTrueRange(5) of data2;
MyEquity= MyInitialBalance+ netprofit + openpositionprofit-commission;
if MM = 2 then myshares= minlist(Intportion((MyEquity*percvol/100)/
MyATR),(marg/100)*Intportion(MyEquity/close));
if MM = 1 then myshares= minlist(Intportion((MyEquity*Myrisk/
Mypercloss)/close),(marg/100)*Intportion(MyEquity/close));
if MM = 0 then myshares= Intportion(MyEquity/close);
if low of data2 <= lowest(l of data2,3) and time > 1000 and time 191
< 1430 then buy myshares shares next bar at (low of data2) limit;

//setexitonclose;
if marketposition > 0 then sell("LXSL") next bar at entryprice
*(1-Mypercloss/100) stop;
if marketposition > 0 then sell("LXTP") next bar at entryprice
*(1+(Mypercloss*2)/100) limit;
I didn’t mention or use a fixed ratio model in the calculations. This model
could certainly be used, but the calculations would be complicated due to the
construction of this particular algorithm. In fact, you’d have to use a lot of
shares to be considered as if it was a futures’ contract and multiply this lot to
calculate the total number of shares to buy each time. You could, of course,
just buy one share, as we did with the futures’ contracts; but, in this case,
the delta would have to be so small it would be an exercise in futility. Due
to these complications, and the potential doubts concerning any advantages,
we won’t use this technique to trade shares.
■ 9.2 Conclusions
The stock market, like any other, is well suited to the application of
position-sizing algorithms. The impact of commissions does, however,
play a significant role, especially in terms of more modest investments. It’s
important to evaluate each model, including the impact of the commissions
in the tests as, on the stock market, they are almost always per trade rather
than proportional to the amount invested.
When permitted, you can also trade with a margin, although this should
be properly coded and taken into account when sizing positions.
192
CHAPTER 10
Portfolio
Management
We’re nearing the end of the book, and it’s important to explore a field
that’s definitely one of the most interesting in a trader’s work. If you’re using
a good strategy and applying correct money management principles to the
193
same produces indubitable benefits, the management of significant assets for
the range of stocks in your basket is certainly something that gives you a
great deal of satisfaction if done correctly.
Portfolio management will assume completely different aspects, depend-
ing on the final client it’s done for.
If you have your own capital and set up a strategy to manage the same,
the development plan can be calibrated to suit your own personality, and the
events that characterize the progress of the equity will be monitored with a
critical, but above all knowledgeable, eye. Managing your capital and plan-
ning your activities to ‘‘control’’ the movements of your assets in a way that’s
sustainable makes it possible to react in an appropriate way when problems
occur. In other words, you’ll have a good idea whether the drawdown you’re
suffering is part of the plan or whether something isn’t going according to
plan and has broken the mold (probably in a definitive way), and it’s better
to cut your losses and call it a day.
Managing your own funds makes you personally responsible for the devel-
opment of the same and, apart from the implications that profits can have
on the family (or more probably not making a profit might have) the only
person you have to answer to is yourself.
If you’re running a professional fund management company, on the other
hand, part of a business setup to manage capital for others, your clients will
be the subscribers of the fund you’re working with. In this case, it will be
a lot harder to ‘justify’ the trends of the capital you’re managing to people
who might not know that much about financial investments, and they cer-
tainly won’t have had a hand in drawing up the investment plan. The classic
example is the different way you’ll see a 40% drawdown. If you’re man-
aging your own capital, a drawdown of that kind might have been part of
the plan, and your risk profile will make such a retracement from the highs
reached easier to withstand. If, on the other hand, a fund managed by a fund
management company suffers such a drawdown, the company will proba-
bly lose a few clients, who might leave with a sour taste in their mouths;
or in any case. at the very least, it would be hard to find new subscribers.
A 40% drawdown, in fact, would be interpreted as the fund management
company’s inability to limit losses, rather than the natural consequence of
an approach that’s not overly conservative and that, in happier times on the
market, would probably produce very interesting results.
As mentioned above, if you work for a company that manages other peo-
194
ple’s capital, you’ll have to plan a strategy that makes the equity line trend
PORTFOLIO MANAGEMENT
as regular as possible.
In fact, while a heavy drawdown can discourage your clients, there’s also
another factor to consider, when trying to keep your subscribers as happy
as possible. If you’re using what’s basically an aggressive system, and it’s a
favourable period, the results will certainly be good, to say the least; and, if
there are no particular difficulties in the following period; but also no major
market fluctuations, there shouldn’t be any particular problems in terms of
drawdown. But, obviously, you won’t see the excellent results obtained in
the previous period either. Once again, your clients might be dissatisfied,
expecting a repeat performance in terms of profits made (or that they read
about in the brochure) during the previous period. You have to consider that
someone who puts their savings in a fund expects to see a trend similar to
the one they read had already occurred in previous periods, and disclaimers
on future trends will be ignored. The client choses a fund and expects it will
continue to perform as it has until then, and as they’ll obviously choose a
fund that performed well up to that point they’ll also obviously expect good
performance in the future.
This is another reason to consolidate the idea that the more regular and
‘gentle’ the trend of the capital is, the better it will look for a professional
fund management company.
The methods used to manage exposure in a stock portfolio, in conceptual
terms, are no different from those we’ve discussed and applied for a single
strategy (or three strategies in Chapter 8).
If you have to monitor several shares, you’ll have signals to enter trades
for some of these stocks at different times or, in rare cases, also at exactly the
same time. The problem for the person managing the capital is, once again,
how much to invest for each single position.
■ 10.1 A Portfolio Approach

In the previous chapters, we discussed futures, and looked at the effect that
different approaches might have on the same. We considered approaches that
quantified the entry points on the basis of the risk the same represented,
therefore establishing parameters for the number of contracts on the basis
of the worst possible event that could occur: the stop-loss.
Other approaches attempted to limit variations in the available capital 195
by measuring the movement of the instrument before entering the trade
and, on the assumption the fluctuations would continue along the same lines
during the trade, the number of contracts on the basis of said variations.
Still other methods started with the minimum possible (one contract for
futures) and increased them in a constant way on the basis of profits made
on the market.
There are pros and cons with each of these approaches, one better suited
than another for the purpose the trader wants to use it for, and also in relation
to the type of market it would be applied to.
When trading on the stock market, we considered the role leverage plays
when choosing the best strategy, and how it can change the limits within
which we can trade.
All these concepts can be adapted to a more consistent basket, and similar
ones can be used, too, based on different logic in line with the same basic
method.
The first step is to refer to Chapter 8, and stress the importance of the
concept of capital to which you refer in each single trade.
The method you use to calculate the equity changes the final result, espe-
cially with systems the positions of which remain open for some time.
So, here once again are the three principal methods:
1. Core equity model. This considers the available capital to be that available in
the portfolio, before opening the position, minus the amounts that would
be lost if the stop-losses of the open positions were triggered. So, this is
quite conservative.
2. Total equity model. This method considers the capital to be the liquidity in
the portfolio, plus the market exchange value of the open positions. In
practice, it allows for the fact that, if an open position is making a lot of
profit, it’s as if the capital has increased even if you haven’t liquidated the
position yet.
3. Reduced total equity model. Like the core equity model, this model considers
the amount of funds before opening positions minus the losses related to
those positions. If stop values are changed, or profits are committed, these
values are added to the capital; it’s as if you consider the position to be
closed at preset values at any time.
After deciding which method to use to calculate the equity, now comes
196
the hard part: choosing the money-management technique.
In the case of a stock portfolio, there are several different methods you
could use, and many are based on very simple assumptions.
Let’s take a look at the most well known and most commonly used
methods.
The simplest method to use by far is equally weighting positions, which
sounds more complicated than the extremely simple principle behind it.
In practice, once you know the number of stocks you intend to monitor,
you divide the available capital into the same number of parts and each part
constitutes the capital used for each stock.
So, if you have 10 stocks to monitor and €1,000,000 available, you
allocate €100,000 to each stock and, when a signal is triggered, you buy
€100,000 of the indicated stock at the exchange value.
The method can develop in two different ways as capital increases. The
first, most obvious way, is to recalculate each time the amount to use for
each single stock. So if the initial €1 million is now €1,050,000, for each
new signal you’d use €105,000, instead of the initial €100,000. In practice,
in the same way as each stock constitutes a fraction of the total number of
stocks you’re monitoring, the capital used for the same constitutes the same
fraction of the total capital of reference.
The second method you could adopt is to continue to use €100,000 for
each signal; once you’ve reached €1,100,000, you’d add another stock to
the basket to make a total of 11.
The first method is certainly the one that lets you manage your portfolio
most effectively, so it grows in a way that’s more in line with the ups and
downs of the market. Using the first method increases exposure on the mar-
ket as your capital increases in proportion to available funds. The second
method, in practice, represents an obstacle to a constant percentage growth
of the capital. In fact, if we consider that it would be quite improbable to be
trading on the market with all 10 stocks monitored at the same time, and
therefore even more improbable that you’d increase your exposure with an
eleventh, there’s no real increase in exposure. This choice, however, allows
greater diversification of the investments. This may mean a higher probabil-
ity of staying on the market, with more stocks for which buy signals could
be triggered, but the level of exposure for each single stock would always be
the same. We would always be using €100,000, and at a certain point in the
growth of the capital, the method might appear to be too conservative.
One method that’s very similar to the previous one is the percent of equity
197
method: in practice, you decide to use a fixed percentage of your capital for
each position.
It’s different from the previous method in that the chosen percentage isn’t
necessarily linked to the number of stocks you’re monitoring. In the example
in which we equally weighted positions, we had 10 stocks and €1,000,000,
using 1/10 or 10% for each stock.
With the percent of equity method we can arbitrarily choose a percent-
age that’s different from 10%, and this could be higher (a more aggressive
method) or lower (a more conservative method). It’s obvious that in this
case, if we used a higher percentage than that represented by the stock in
relation to the total number of stocks monitored, we could find ourselves in
a situation in which we couldn’t make a trade because we didn’t have suffi-
cient available capital. If, for example, with €1,000,000 we decided to use
a percent of equity method with a percentage of 20%, and we found our-
selves contemporarily with open positions for 5 of the 10 stocks monitored,
we couldn’t open a trade for a signal that would have triggered a sixth stock,
as all our capital is tied up on the market (5 stocks using 20% of the capital
each makes a total of 100% of the capital).
In any case, it must be said that the percent of equity method is one of the
most used methods: it’s extremely simple as you always know how much
capital you have available. (Remember, though, that this does in any case
depend on the initial choice you made on how you’ll consider the capital; in
other words, in terms of core equity, total equity or reduced total equity.)
All you have to do is calculate the percentage chosen and buy the appropriate
number of stocks for the resulting amount.
The two methods described give you an idea of how to ‘distribute’ avail-
able funds and that it’s not a good idea to stay too long with a single stock.
Thinking initially though in terms of capital held, and adjusting the num-
ber of shares bought only on the basis of this parameter, there’s the risk of
creating an imbalance in your portfolio.
It’s certainly important to consider the consequences of an investment
and not just consider things in terms of before (the capital held), but also in
terms of after (the trend of the trade).
One way to follow the evolution of an entry into the market is, as we saw
in previous chapters, to consider how much this investment could lose and
decide beforehand to limit the global effect of said negative event.
You simply decide beforehand what part of the capital you are willing
to risk if you close the trade with a loss; and, in consideration of the open
198
position risks, you adjust the relevant number of shares as a consequence.
This method is better known as percent risk as it’s a percentage risk on which
the your choice is based.
So, if you have €1,000,000 and don’t want to risk more than 2% of the
entire capital, you’d analyze the worst event for the position you intend to
open. If, for example, the position has a 4% stop-loss, you’ll see we can
buy using as much as €500,000, which is half the available capital; in fact,
if a stop-loss is triggered half of the capital would lose 4%, which would
obviously be 2% of the total capital.
Even with this straightforward example, it’s easy to see that, when
using this type of approach, you might easily not be in a position to make some
trades as you won’t have enough capital (in the above example, in fact, we
used half the capital to enter one trade, and all it would take would be another
similar trade to have committed all our available capital on the market).
It’s true that this example is only used to illustrate the method. In fact,
when using this method the risk percentages are very low, in order to create
the least possible disturbance to the trend of the equity line and allow for
greater diversification, always on the condition you have capital available.
The first methods described, as mentioned above, don’t consider the con-
sequences of the trade but just the capital. Percent risk attempts to calibrate
entering a trade in terms of the worst event that could occur. It’s true that,
if we adopt the same strategy for all the stocks in the basket in question, and
this strategy had a 5% stop-loss, for example, once again all the stocks would
be considered in the same way. This would certainly give you a good deal of
control over the equity trend in terms of loss containment, but the exposure
would be practically the same for each single stock. If we risked 0.5% of the
capital and the strategy had a 5% stop-loss, this would mean we would risk
€5,000 of the €1,000,000. So this €5,000 would be 5% of the capital allo-
cated to each single trade, as this is how much we would lose if the stop-loss
was triggered. The stock would therefore be bought for €100,000. In fact, if
the 5% stop-loss is triggered with €100,000, we would lose €5,000, which
is 0.5% of the entire capital.
So it’s clear that you don’t start with the available capital but with the
losses, and this is positive, but you continue to treat each stock in the same
way, the strategy is the same and the capital allocated to each trade is the
same regardless of which stock is chosen. This is obviously true in the case
in which the strategy provided for a percentage stop-loss, as in the above
example. 199
Once again, there’s the risk that the portfolio in question wouldn’t be
balanced in the best possible way.
If we take a banal example, it’s easy to see that holding €10,000 in ENEL
isn’t the same thing as holding €10,000 in Tiscali. Volatile stocks cause more
disruption in the equity line than ‘quiet’ stocks.
So, in order to effectively deal with this, let’s once again consider the
volatility of stocks, just as we did with the percent volatility model in
Chapter 4.
As we saw in that chapter, the volatility of stock can be measured by cal-
culating an average of the fluctuations of said stock during the day. As is well
known, the difference between the day’s high and low is called the range, and
the bigger it is, the more volatile the stock is.
True range was introduced to replace range in order to provide a more
accurate figure for the multiday trend, allowing for the possibility that the
previous day’s close lies outside the daily range and, if applicable, it shows
the distance of the same from the bar high or low to the range value (see
Figures 4.18 and 4.19).
In Chapter 4, we considered an average of five days for the true range.
It must be said that for medium- to long-term strategies it’s preferable to
consider longer periods. We said that 20 days serve as a good compromise
(on daily charts, this is approximately the equivalent of a one trading month).
Therefore, calculating an average at 20 days (or another period, depending
on the type of system chosen) of the true range gives you an idea of the
average stock fluctuation. This fluctuation is measured in terms of absolute
value, and may, for example, be 1.5, which means that the stock in question
on average moves 1.5 points during the day (for example, from 30 to 31.5).
In this case, of the stocks in our list and standard trading, this value also
corresponds to the value in euros. If, in fact, a stock moves from 30 to 31.5,
we would say it moved €1.5. What we want to do is measure the volatility
value in euros in order to evaluate the effect on the capital. With futures it’s
different, as the volatility in euros is usually not the same as the movement
in points. With Italian futures, for example, each point is worth €5, and a
true range average of 500 points is the equivalent of a volatility of €2,500.
Returning to stocks in order to simplify things, we’ve seen that one stock
has an average volatility of €1.5 (in the above, example). If we buy 1,000,
we’ll have an average of €1,500; buying 2,000 will be €3,000, and so on.
200 Now let’s consider our available capital to be €1,000,000. We deem the
possibility of suffering variations of over 0.5% to be too onerous; in other
words, if there were fluctuations of over €5,000 in the open positions, at this
point, with our stock that has an average volatility of 1.5 points we can buy
3,333 shares. So:
5,000∕1.5 = 3,333.333.
If we buy more, we run the risk of suffering greater fluctuations than the
chosen €5,000.
The volatility of the stock can be used in two ways that translate into the
same number of different money management methods. The first of the two
is the percent risk with ATR method, which practically uses the same steps as
the percent risk method but, instead of using the stop-loss to limit negative
events, it uses a multiple of the average true range (ATR) as the limit value
for the fluctuation.
If, for example, we take the 10-day ATR as a significant volatility value,
we might consider a fluctuation that’s twice said value to be a significant level
at which to limit the trade. At this point, you establish the percentage of the
capital you’re willing to let fluctuate to this limit and calculate the number
of stocks to buy on the basis of this figure.
Let’s take, for example, the Canistracci Oil stock, which has a 10-day ATR
of two points (the choice of this number is also arbitrary and in general is
linked in different ways to different strategies). Supposing we have an avail-
able capital of €100,000 and we don’t want to expose more than 1% of said
capital to the risk of excessive portfolio fluctuation. This means we want to
prevent the portfolio losing more than €1,000 during the trade we’re going
to open, considering that we want to avoid this event in cases of excep-
tional volatility, so we multiply the 10-day ATR by 2 and take this value as
a reference. The result is volatility in exceptional cases of 4 points (2 mea-
sured multiplied by 2). We don’t want this to produce losses of over €1,000,
so it’s easy to calculate 1,000/4 = 250, which is the maximum number of
Canistracci Oil shares we can buy.
Often, this method is combined directly with the basic strategy in which,
instead of a percentage stop-loss, you use a stop level or a trailing stop equal
to a multiple of the ATR. In this case, the value of the stop in the strategy cor-
responds to a multiple of the measured volatility of the instrument (which
is also logical if we consider it’s better to close a position if the instrument
starts moving in the opposite direction). 201
The second method that uses volatility is the actual percent volatility model.
In this case, you don’t consider the system stop-loss but simply its ATR – let’s
say, over 20 days. This value is used as the volatility of reference for the
instrument. Now, if we establish how much we’re willing to let our total
capital fluctuate, we can calculate the number of shares to buy on the basis
of the ratio of the two values.
If the 20-day ATR of the above Canistracci Oil stock was 1.8 points and we
decided that for our capital of €100,000 we didn’t want to suffer fluctuations
of over 1%, this means we mustn’t buy more than:
1
100,000 ∗
100 = 555.55,
1.8
so we should buy no more than 555 shares.
The above lays the foundations for correct portfolio management, but as
always, it’s important to take a look at some results to fully appreciate the
different ways the equity line can behave depending on the method chosen.
In Chapter 8, we saw how combining different strategies can produce
astonishing results. In that chapter, we limited the trading to three futures,
adopting the same strategy for all.
In this case, however, we’ll study a stock portfolio, and to simplify things
we’ll limit ourselves to only 10 stocks as an example, in consideration of the
fact that the principles described can easily be applied also to more consistent
baskets.
Once again, we’re interested in the trends produced by different money
management techniques, so we’ll take the case of just one single strategy
applied in the same way to all the stocks.
We’ll compare the results using the same percentage of capital for each
stock with that to using percent volatility and see which approach limits
drawdown most effectively in relation to profit made.
First, let’s take a look at the rules of the strategy we’ll be using to trade
both long and short on daily bars.
We’ll use a faster moving average with 20 periods and a slower moving
average with 45 periods, as well as a trend reference average of 200 periods.
The long positions will only be opened if the last close is above the 200
periods moving average and the short positions will only be opened if the
last close is below the same average.
202
The long trades will be opened if a day closes with the faster moving
average that crosses over the slower moving average; vice versa, the short
trades will be opened when the same averages cross in the opposite way.
The positions will be closed when the averages cross in the opposite way
or when the trend phase changes (in other words, long positions will be
closed if the faster moving average drops below the 200 period average and
short positions will be closed if the faster moving average rises above the 200
period average on an uptrend).
The code for this strategy follows, with the option to choose between
percent volatility and trading a percentage of the equity.
Input: Slow(20), fast(45), percvol(2), eqperc(10);

var: myshares(1000), MyATR(1);
myshares = intportion(Portfolio_Equity*(eqperc/100))/close;
MyAtr = averagetruerange(slow);
if percvol > 0 then myshares = intportion((Portfolio_Equity*percvol/
100)/MyAtr);
if average(c, fast) crosses above average(c, slow) and c > aver-
age(c,200) then buy myshares shares next bar at market;
if average(c, fast) crosses below average(c, slow) and c < aver-
age(c,200) then sellshort myshares shares next bar at market;
if average(c, fast) crosses above minlist(average(c, slow), aver-
age(c,200)) then buytocover next bar at market;
if average(c, fast) crosses below maxlist(average(c, slow), aver-
age(c,200)) then sell next bar at market;
With reference to Chapter 9, we’ll study the results complete with a com-
mission of €10, as used since 2001 (beginning of 2018).
We’ll trade with an initial capital of €1,000,000 euro and a portfolio of
10 stocks listed on the Italian market: 203
A2A
ENEL
ENI
ERG
G
MB
MS
TIT
AGL
STM
Figure 10.1 shows the result when entering the trade with 10% of the
total equity for each position.
FIGURE 10.1 Performance report always allocating 10% of the total equity per
position.
1,600,000
1,400,000
Equity (€)
1,200,000
1,000,000
204
Drawdown (€)
–100,000
–200,000
Drawdown (%)
–10
–20
3/31/2005 7/24/2009 6/25/2013 6/6/2017

Date
FIGURE 10.2 Equity line always allocating 10% of the total equity per position.
Figure 10.2 shows the corresponding equity line and Figure 10.3 shows
the distribution of profits between the various stocks. Finally, Figure 10.4
shows the annual distribution of profits.
FIGURE 10.3 Distribution of profits between the stocks, always allocating 10% of the
total equity per position.
The final result is a profit of approximately €650,000 with a maximum

percentage drawdown of around 24.5%.
If we wanted to compare similar profits, we’d have to use a percent volatil- 205
ity model that aims to limit daily fluctuations to within 0.3 (as can be seen
in the code, we used the same faster-moving average period to calculate
the ATR).
The results are shown in Figures 10.5–10.8.
As can be seen, there’s not much difference, although there is a slight
advantage with the percent volatility model. This, in fact, makes slightly
more profit while suffering a few percentage points less drawdown. The
drawdown curve also appears better in Figure 10.6 than in 10.2, while there
aren’t any substantial changes in the distribution of profits, both in terms of
stocks and over the years.
Personally, I prefer the percent volatility model, but as it’s harder to cal-
culate the data to enter the trade, others might be more prone to choose
the simpler percentage allocation model. If everything is automated, then of
course the choice would be percent volatility.
206
FIGURE 10.4 Annual distribution of profits, always allocating 10% of the total equity
per position.
FIGURE 10.5 Performance report with a percent volatility model of 0.3%.

1,600,000
1,400,000
Equity (€)
1,200,000
1,000,000
Drawdown (€)
–100,000
–200,000
Drawdown (%)
–10
–20
3/31/2005 7/24/2009 6/25/2013 6/6/2017
Date
FIGURE 10.6 Equity line with a percent volatility model of 0.3%.
207
FIGURE 10.7 Distribution of profits between the stocks with a percent volatility
model of 0.3%.
208
FIGURE 10.8 Annual distribution of profits with a percent volatility model of 0.3%.
■ 10.2 Some Improvements to the System

For a moment let’s leave position sizing concepts aside and add some simple
filters to the entry points to make them cleaner and avoid the false signals
that can always be found in this type of strategy.
With this type of strategy the impulse that derives from the setup related
to the moving average crossover is very important. In this case, there are
a variety of different dynamics between long and short moves as they are
related to the very nature of the stock market, which moves in a very dif-
ferent way depending on whether it’s on an uptrend or downtrend. Even in
terms of acceleration of the movements, they often behave in opposite ways.
So, on the basis of these considerations, it may be expedient to filter long
entry points only when the prices show a certain propensity for accelera-
tion, while short entry points should be filtered when the price is excessively
dynamic.
We can measure acceleration using a well-known indicator in technical
analysis: the ADX. For our purpose, we can use its value over a period of
5 days and filter long entries only if it is above 35 and short entries only if
under 55.
These values were chosen as a result of some tests not here included to
avoid departing from the main subject of discussion of the book, but they are
in any case indicative values, and similar effects can be obtained with values
other than those mentioned: such as, for example, a threshold of 30 for long
entry points and 45–50 for short entry points.
The code of the system complete with filters, follows.
Input: Slow(20), fast(45), percvol(2), eqperc(10);

var: myshares(1000), MyATR(1), myEntries(0);
myshares = intportion(Portfolio_Equity*(eqperc/100))/close; 209
MyEntries = intportion(100/eqperc);
MyAtr = averagetruerange(slow);
if percvol > 0 then myshares = intportion((Portfolio_Equity*percvol/
100)/MyAtr);
if average(c, fast) crosses above average(c, slow) and c > aver-
age(c,200) and adx(5) > 35 then buy myshares shares next bar
at market;
if average(c, fast) crosses below average(c, slow) and c < aver-
age(c,200) and adx(5) < 55 then sellshort myshares shares next
bar at market;
if average(c, fast) crosses above minlist(average(c, slow), aver-
age(c,200)) then buytocover next bar at market;
if average(c, fast) crosses below maxlist(average(c, slow), aver-
age(c,200)) then sell next bar at market;
Let’s take a look at the results after applying the filters, also to see whether
they’re useful or not. Figure 10.9 shows the performance report of the sys-
tem with the filters using 10% position sizing on the equity followed by the
equity line in Figure 10.10, the distribution of stocks in Figure 10.11, and
the annual distribution in Figure 10.12.
There’s a notable improvement on all fronts, not only has the net profit
increased, the drawdown has also decreased considerably. The trend of the
210 FIGURE 10.9 Performance report with ADX filter and 10% exposure per position.
1,750,000
1,500,000
Equity (€)
1,250,000
1,000,000
Drawdown (€)
–100,000
–200,000
Drawdown (%)
–5
–10
–15
3/31/2005 7/24/2009 6/25/2013 6/6/2017
Date
FIGURE 10.10 Equity line with ADX filter and 10% exposure per position.
FIGURE 10.11 Distribution of profits between the stocks with ADX filter and 10%
exposure per position.
211
FIGURE 10.12 Annual distribution of profits with ADX filter and 10% exposure per
position.
equity line is better and the number of losing stocks has decreased as has the
number of years that ended with a loss. This is without a doubt an apprecia-
ble result. Let’s see what sort of figures we’d have obtained using the 0.3%
percent volatility model (Figures 10.13–10.16).
Also in this case the results are quite similar, with a slight improvement
in drawdown compared to the equal weighting of positions in percentage
terms.
212 FIGURE 10.13 Performance report with ADX filter and 0.3% percent volatility.
1,750,000
1,500,000
Equity (€)
1,250,000
1,000,000
Drawdown (€)
–100,000
–200,000
Drawdown (%)
–5
–10
3/31/2005 7/24/2009 6/25/2013 6/6/2017

Date
FIGURE 10.14 Equity line with ADX filter and 0.3% percent volatility.
FIGURE 10.15 Distribution of profits between the stocks with ADX filter and 0.3%
percent volatility.
213
FIGURE 10.16 Annual distribution of profits with ADX filter and 0.3% percent
volatility.
TABLE 10.1 Summary of results.
System Net Profit Max. % Drawdown
10% Equity no filters 644,894.76 24.48%

0.3% Perc Vol no filters 683,150.71 22.01%
10% Equity with ADX filter 877,773.58 16.31%
0.3% Perc Vol with ADX filter 876,633.01 14.29%
Table 10.1 is a summary of the results to give you a complete picture of

the situation.
■ 10.3 Conclusions
In Chapter 8 we saw how using several instruments simultaneously made it
possible to multiply the effects of money management strategies, above all
with the aim of increasing the initial capital. If we start trading with several
instruments and try to make the equity line softer, the techniques we’ve
studied remain valid even when applied with lower risk percentages. It’s
214 always important to look for the ideal compromise between an increase in
profits and surviving negative periods unscathed. If you’re accountable to
others who’ll also be watching the trend of the capital, you should try to
make this trend as stable as possible, to leave less room to doubt your work
in the minds of those who know less about these sorts of transactions.
The systems used, as always, aim to limit fluctuations in the capital, both
after closing a trade, and therefore when calculating entry points in terms of
the maximum possible loss, or with the open trade, when calculating entry
points on the basis of the volatility at that particular time.
For every system you intend to use, you should find the money manage-
ment technique that’s best suited for your own particular disposition to risk
and how thick-skinned you are in terms of withstanding the ups and downs
of the market.
Sometimes, as well as the by-now well-known money management tech-
niques, it can be expedient to consider the type of signals used and try filters
that can improve the quality of the trades, which, with position sizing mod-
els, can produce notable improvements in the final scenario.
CHAPTER 11
Discretionary
Trading
■ 11.1 Trading Criteria and Definition
After reading the previous chapter, you’ll probably be thinking the money
management techniques in it must be combined with a trading system. 215
Certainly, in order to plan things in an effective way, you should simulate
the strategy you’re thinking of using to see how it behaves, so it’s always
a good idea to draw up a trading system before you start trading on the
market. But it isn’t absolutely necessary to have one to apply the rules of
money management. The only thing that is necessary to do a good job is to
have a clear idea of what you want to do.
Unfortunately, people often start out without knowing what to do to
begin working seriously, and start improvising as if this is the normal way
to go about things. So, what does having a clear idea mean? And what does
it mean if you don’t have a clear idea?
First and foremost, let’s define discretionary trading as it will be con-
sidered in this chapter: discretionary trading is taken to mean trades that
aren’t made on the basis of signals from a trading system but as a result of
observing the market, either on chart or even on the order book. We won’t
consider scalping as, with this type of trading, which is quite fast and fre-
netic, the application of money management principles tends to be ‘lost’
in consideration of the potential absorption of the amounts traded on the
market. Only a meticulous and honest study of your scalping can rationalise
the amounts traded in an attempt to apply the rules of money management.
We will consider longer trades, however, which may be intraday trades
with a profit target or trades closed end of day, or long-term multiday trades
based on observations of daily charts, or trades made based on accumulation
seen on the order book.
All too often, unfortunately, one studies potential market entry points
meticulously, while leaving exit points to chance. The typical behaviour of a
discretionary trader (if the reader is a discretionary trader but can’t identify
with this profile, all the better) is to look for ‘good’ entry levels, choosing
a proper size (in other words, the number of shares to be traded) and the
actual trade to enter the market. It’s only after the trade has been opened
that the trader starts to think about how and where to exit; it’s in this phase
that most of the troubles arise. If the stock falls, people often exit too soon
on the basis of the ‘cut your losses and run’ rule, or stop levels are set further
and further away, studying charts with a wide variety of time frames in order
to find a good reason (or good excuse) for not closing a losing trade.
Appraisals of this kind, influenced by feelings based on an idea of how
the trade is going, are often detrimental and can destroy all the good work
done up to that point. The only justification for continuously following a
216
trade is if the trade was opened on the basis of the meticulous observation
DISCRETIONARY TRADING
of buy and sell movements, in order to be able to exit if things start going
in the opposite direction to what was planned when it was opened. Trades
of this kind, however, require talent and exceptional skill and, while some
may think they have such things, those who trade like this don’t last long on
the markets.
To manage your trading properly, all your considerations should be made
beforehand. Entry levels are definitely important, but exit levels are just as
important, whether to take a profit or when a stop-loss is triggered. So the
whole trade should be planned before it’s opened and when entering the
market you must already have a clear idea of the levels you’ll exit at.
Above all, the exit level if things go wrong must be established right from
the start, in this way you have a clear idea of the risk involved when opening
the trade.
If you have a clear idea of the entry and exit levels, you can apply money
management principles with due consideration.
In the previous chapters we saw many times how simulations based on the
past can be used to establish which money management model is best suited
to the trading system used. But now, in the case of discretionary trading
there’s no historical data from any trading system to base decisions on. Even
if a particularly well-organised trader kept note of past trades, it would be
quite difficult to have a sufficiently vast archive to base valid appraisals on.
So, decisions must be taken on the basis of more generic appraisals, which
have an absolute value, however.
First, it can be said that, if a trader is trading stocks (with or without
leverage) rather than futures, I wouldn’t recommend using the fixed ratio
method. As we saw, this method is difficult (or rather, complicated) to apply
to trading on the stock market, and if it’s complicated when you can base
your decisions on a sufficiently high number of historical trades, it will be
even more so without historical data to base decisions on.
So, when trading stocks and shares, the best thing is to choose between
the fixed fractional method and the percent volatility model.
The choice between these two methods should be made on the basis of
the types of trade you’re going to open, and the characteristics of the type
of exit to use for the same.
In previous chapters I explained that the percent volatility model is
better for a very tight stop-loss, and in fact using the fixed fractional method
in these cases meant the risk of trading large positions with numerous
217
consecutive losses. The fixed fractional method, as you’ll recall, considered
the potential loss and on the basis of the same calculated the number of
contracts (or shares) in order to limit the total percentage impact on the
capital. When using a very tight stop-loss, the loss will be quite limited
and, for the same percentage impact, you can enter the trade with more
contracts. For example, if you have €100,000 and enter a trade using a
stop-loss at 1 percentage point from the entry point and decided you didn’t
want to risk more than 0.5% of the total capital on each trade, you can
open the position at €50,000. If the stop of the position is tighter, and let’s
make a hypothesis that it’s the same 0.5%, you can open the position using
the entire capital of €100,000. Actually, if you wanted to use an extremely
tight stop-loss of 0.25% (which might occur when buying on a support
level and deciding to use a stop just below the value of said support) you
could even enter the market with €200,000 (which would obviously only
be possible if you were using leverage).
The percent volatility model on the other hand doesn’t consider the
potential loss, but measures the movements of the stock and calibrates the
entry based on the Average True Range of the last period.
If you want to use this method, you will have to have already calculated
the average at n days of the true range for the stock. As this value is calcu-
lated on the basis of daily charts, you could calculate it the previous evening
and therefore also know how many shares you want to trade with.
■ 11.2 An Example: Mediaset

Let’s take a look at an example to see what might happen in practice.
We’ll trade Mediaset stock using the daily chart. In the beginning of
November we have a setup to place a trade. After a long down trend the
stock rebounds and settles at 9.275 to then fall again but without breaking
through the lows of the long down trend. A congestion channel follows. One
hypothesis for the trade might be to enter long when there is a breakout at
9.275, with a 8.78 stop-loss below the lows of the long down trend.
A hypothesis for the target could be 9.90, which was a significant level in
the month of August.
As this is expected to a be a long-term trade we’ll use precise levels but
decide to enter at 9.30 with a 8.75 stop, while still using 9.90 as a target.
Figure 11.1 shows the chart with details of the above.
218 There’s a long wait for the trade. In general, this is a good sign: the stock
will stay in the congestion channel longer. In any case the trade would have
been entered on 5 January 2006 and the target would have been reached on
6 February. As the stock reached exactly the target high of 9.90 that day it
is not certain the sell order would have been executed, but on 7 February it
passed 9.90 and the trade can be considered closed.
The point we want to make, however, isn’t the description of the trade,
but rather how to calculate the number of contracts we would have bought
on 5 January.
We’re not using a very tight stop and we can consider using a fixed frac-
tional method without qualms. Let’s assume we don’t want to risk more than
2% of our capital, which is €100,000.
The first thing to do is evaluate the stop-loss in percentage terms. As
mentioned above, the stop was set at 8.75 and the entry at 9.30, so:
9.30 − 8.75 0.55
stop percentage = ∗ 100 = ∗ 100 = 5.91%
9.30 9.30
MS.MI LAST-Daily 02/07/2006 C=9.840 –.030 –.030% O=9.900 H=9.920 L=9.830 V=2320057
10.400
10.200
10.000
9,90 9.800
9.600
BUY
9.400
9,275
9.200
9.000
8.800
8,78
Average True Range(5) 0.1930
0.2200
0.1800
0.1400
0.1000
Jul Aug Sep Oct Nov Dec 06 Feb
FIGURE 11.1 An example of a discretionary trade with Mediaset stock.

Now let’s calculate the amount we’re willing to risk. As mentioned above
our percentage risk is 2%, so:
amount risked = 100,000 ∗ 2∕100 = €2,000.
With these two parameters, we can now calculate the amount to invest in
the trade. The capital used for the trade must be equal to an amount 5.91%
of which is €2,000 because, if the stop is triggered, losing 5.91%, we don’t
want to lose more than €2,000.
investedcapital ∗ 5.91
= 2,000
100
2,000 ∗ 100
investedcapital = = 33,841.
5.91
This means we can enter a Mediaset trade with €33,841 because, if we
lose 5.91% of the same, it won’t be more than a 2% loss of the total available
capital (€100,000).
Let’s suppose we enter the trade at 9.30, which means:
number of contracts = 33,841∕9.30 = 3,638.
To check the calculations, let’s see what would happen if the stop is trig-
gered for those 3,638 contracts, we’d lose:
220
(9.30 − 8.75) ∗ 3,638 = 0.55 ∗ 3,638 = 2,000.9.
The figures add up!

Here’s a formula you can use to directly calculate the number of contracts
on the basis of the entry price and stop if used, as well as obviously the risk
chosen and available capital:
investedcapital ∗ risk%
number of contracts =
(entrylevel − stoplevel) ∗ 100;
in our case:
100,000 ∗ 2
number of contracts = = 3,636.
(9.30 − 8.75) ∗ 100
Note that the difference between the two calculations (3,638 compared
to 3,636) is only due to the subsequent rounding off in the first case; this
obviously doesn’t have a significant effect on the final result.
This formula is valid for stocks but not for futures. For the latter, you
should consider the value of the ticks for a precise appraisal, but this isn’t
what we’re looking at in this case.
The trade would have closed with the following result (trading 3,636 con-
tracts as in the direct formula):
gain = (9.90 − 9.30) ∗ 3,636 = €2,181.6.
Someone might say it wasn’t a very good trade because, as shown by the
calculations, it produced a profit that was quite similar to the potential loss.
These kind of objections, however, should be considered more in terms of
risk management as, it must be said again, money management establishes
how much to trade, but not how. Anyway, if we want to make a case for this
choice, note that although the risk/benefit ratio is more or less equal in
monetary terms, for this type of entry there’s a higher probability of closing
with a profit than at a loss due to the congestion in the period before the
breakout.
How might we have used the percent volatility model?
The chart also shows the five-day ATR. The five-day range was chosen
as this represents a good compromise for medium-term trades like the one
we’re going to open using this discretionary approach (this average is calcu-
lated on the basis of the distance of the target prices from the entry price,
and obviously no one knows in advance for how long the position will be
open).
Remember that in Chapter 4, and in other examples, we saw that the 221
percent volatility model limits fluctuations in the total capital, and the per-
centages used in this model were also generally lower than those of the
reduced percentage method. So let’s make a hypothesis that we want to trade
limiting fluctuations in the capital to 0.5% of the total.
Every morning we’ll check the value of the ATR indicator and on 4 Jan-
uary (obviously we’d have the figure of the previous day on the morning of
5 January, and use this) it was 0.1230, so now let’s take a look at how we’d
proceed:
First, we’d calculate the admissible range:
range = 100,000 ∗ 0.5∕100 = 500.
On the basis of our observation we know that the volatility of 1 MS share
is €0.1230; so, in order to remain within the limit of a max. range of €500
we could buy:
number of contracts = 500∕0.1230 = 4,065.
As you can see, the number is quite close to that calculated using the
procedure that limited the risk if the stop-loss was triggered.
Also in this case, there’s a formula that can be used:
totalcapital ∗ extension%
number of contracts =
AverageTruerange(n) ∗ 100
where n is the number of days for which the ATR is calculated.
In this case, the trade would have made:
gain = (9.90 − 9.30) ∗ 4,065 = €2,439.
I’ll repeat a concept I already expressed in the previous chapter: the fixed
fractional method is used to limit the effect on the capital if the trade is
stopped out, so when closing the trade: in fact, you calculate and limit the
effect if a stop-loss is triggered. The percent volatility model, however, limits
the effect during the trade. In other words, you calculate the volatility of the
stock in the previous days to see what effect it might have on the equity line
while you have an open position.
Note that the percent volatility model doesn’t in fact consider the strate-
gic levels of the trade you are going to open. The formula doesn’t consider
the entry level (the number of contracts is calculated without even consider-
ing the entry price) or where the stop-loss is positioned, and it only considers
how much the stock moved in the latest period, so how much it could move
while we have a position open.
222
If the stop is a lot tighter, the percent volatility model calculation would
produce exactly the same result (if applied to the same day obviously), while
the fixed fractional method calculation would change, a lot.
For example, if instead of choosing the low of the long down trend we
chose the low of the previous bar as the stop-loss level, the figure to monitor
would be 9.06 (low of 4 January 2006), and in this case we would have
entered the trade with:
100,000 ∗ 2
(9.30 − 9.06) ∗ 100
This is a very different result, which would, in this case, have produced
more profit, as the stop was never triggered in the following bars.
This doesn’t mean a choice of this kind is better, but goes to prove the
intrinsic difference between the two methods. What’s more, we applied
the percent volatility model using 0.5%, which on the stock market may
be suitable, but if we’re trading futures with this percentage, we’d run the
risk of not being able to make many trades (we’ll go into this in greater detail
later), and it would be better to use higher percentages better suited to the
characteristics of the futures’ market with the relevant leverage.
Once the trade has been opened, you follow its evolution as planned.
If you want to open a second trade, you’d do so on the basis of the same
considerations you made for the first, with the only difference being that
the capital you’d base your calculations on wouldn’t be the same. The rules
presented for the core equity model, total equity model, and the reduced
total equity model apply when considering the capital.
With the core equity model the capital to be considered for the following
trade would be €98,000 when using percent risk (in fact, you deduct the
stop-loss for the open position equal to €2,000 from the available capital), or
approximately €97,764 when using percent volatility (with 4,065 contracts
the loss with an 8.75 exit would be approximately €2,236).
The total equity model, each time, considers the value of the open posi-
tion and adds this to the liquidity held.
If, for example, we bought 4,065 contracts (as we did with the percent
volatility model), we’d have used:
4,065 ∗ 9.3 = €37,804.5.
So we’d have a liquidity of:
100,000 − 37,804.5 = €62,195.5.
(We’re not considering the cost of the commissions in order to simplify 223
things, but you should always bear in mind that the value of your liquidity is
usually clearly shown on the platform you’re trading with so you don’t have
to calculate it.)
If the new position was opened on 26 January, we could refer to the exist-
ing position by calculating the value with a 25 January close. On this date,
the MS stock closed at 9.37 so the value of the open position is:
4,065 ∗ 9.37 = 38,089.05,
and our total equity would be:
62,195.5 + 38,089.05 = 100,284.55.
This is the capital to use in the total equity model calculations.
With the reduced total equity model, once again we’d consider 98,000 (or
97,764) as the capital, but supposing that once the MS stock passed €9.47,
it would be expedient to move the stop to 9.11. See Figure 11.2.
At this point, if the stop is triggered, we’d risk losing:
4,065 ∗ (9.3 − 9.11) = €772.35.
MS.MI LAST-Daily 02/07/2006 C=9.830 –.020 –0.20% O=9.900 H=9.920 L=9.830 V=1583478
10.400
10.200
10.000
9,90
9.800
9.600
9,47
9.400
9,275
9.200
9,11 9.000
8.800
8,78
0.2200
0.1800
0.1400
0.1000
Nov Dec 06 Feb
FIGURE 11.2 Moving the stop of the position to 9.11 after the high of 9.47.
The new capital of reference would therefore be:
100,000 − 772.35 = 99,227.65.
After closing the position, whatever the method used, the capital of ref-
erence would be (if there are no other open positions):
100,000 + (9.90 − 9.30) ∗ 4,065 = 102,439.
■ 11.3 Adjusting Volatility During the Trade

Using the percent volatility model from the start the idea is to limit
fluctuations in the capital with an open trade. On this basis, we can calculate
a certain number of shares (or Futures’ contracts) to buy when opening the
trade, in terms of the volatility in the previous period.
But if the trade remains open for some time, the volatility can change
considerably, and the movements of the equity line may no longer be within
the preset limits.
If the volatility decreases, this isn’t a problem, but if it increases, this could
cause some difficulties.
On the basis of the above line of reasoning one might periodically review
the number of contracts in order to keep the fluctuations in the equity line 225
more or less constant. In theory this could be done every day, but doing so
would be very time consuming. We could make a hypothesis of checking
things every five trading days (once a week) and adjust the situation to the
new volatility parameters.
In practice, every five days you note the new ATR and check how many
contracts you should have on the basis of this figure. Then decide whether
to adjust your portfolio.
Let’s take a look at the ATR figures during the MS trade, starting 4 January
and checking every week (Table 11.1).
We’ll proceed week by week. As mentioned above, on 5 January we enter
a trade with a position of 4,065 contracts. At the day’s close on 11 January,
using the total equity model our capital of reference would be:
capital of reference11 Jan = 100,000 − 4,065 ∗ 9.3 + 4,065 ∗ 9.355
= 100,223.575.
The number of contracts that could be used for said capital of reference,
with a maximum expansion limited to 0.5% as in our case and an ATR of
TABLE 11.1
date ATR(5) Close of day Open next day
4/1/2006 0.1230 9.22 9.25

11/1/2006 0.1320 9 9.36
18/1/06 0.1540 9.3 9.33
25/1/06 0.1500 9.37 9
1/2/2006 0.1540 10 9.78
0.1320 (the value on 11 January) would be:

100, 223.575 ∗ 0.5
0.1320 ∗ 100
It’s clear that, at this particular time, we’re overexposed: we have more
contracts that the recommended number. So we decide to sell the excess
contracts when trading opens that day, and will sell:
4,065 − 3,796 = 269 contracts.
In its own way, this method made us save part of the profit made; but it
could also have made us sell at a loss if the market prices were lower than
226
our entry level and the volatility just increased.
On the evening of 18 January the sums required to calculate the capital of

reference become a little more complicated as some of the contracts were
sold (in practice, it’s a lot simpler as the trading platform shows both the
value of the open position and the liquidity of the account).
+ 269 ∗ 9.36 = 100,016.14.
Remember the calculation is based on the initial capital – the initial
amount bought (initial number of contracts multiplied by the entry price)
+ the current value of the portfolio (current contracts in the portfolio
multiplied by the close price on 18 January) + contracts sold for the sale
price (opening price on 12 January).
Using these figures and an ATR of 0.1540, we obtain:
100, 016.14 ∗ 0.5
0.1540 ∗ 100
Once again (if we want to use this approach), we need to sell part of the
contracts in the portfolio. This time:
3,796 − 3,247 = 549 contracts.
These 549 contracts will be sold when trading opens next day at 9.33.
Let’s take a look at next week, first calculating the capital of reference:
+ 269 ∗ 9.36 + 549 ∗ 9.33 = 100,259.9.
The ATR is 0.1500, so:
100, 259.9 ∗ 0.5
0.1500 ∗ 100
This time the system tells us we can increase the number of contracts. We
could also not increase the number of contracts, simply limiting ourselves to
stay within the set range, but to complete the analysis, let’s say when trading
opened next day (at 9.405) we bought:
3,341 − 3,247 = 94 contracts.
At the beginning of February, we repeat the procedure and calculate the
capital of reference:
227
capital of reference1 Feb = 100,000 − 4,065 ∗ 9.3 + 3,341 ∗ 9.775
+ 269 ∗ 9.36 + 549 ∗ 9.33 − 94 ∗ 9.405
= 101,609.72.
Once again remember that, of this figure, the open position amounts to
3,341 * 9.775, while all the other figures are part of the liquidity calculation
as they are closed trades.
The ATR on 1 February is 0.1540, so we calculate:
101, 609.72 ∗ 0.5
0.1540 ∗ 100
Therefore, we have 42 contracts too many, which we sell when trading
opens on 2 February at a price of 9.78.
No further adjustments are necessary, as the trade will be closed at the
target of 9.90 before another week has passed.
The final calculation of the result is:
gain = (−4,065 ∗ 9.3 + 269 ∗ 9.36 + 549 ∗ 9.33 − 94 ∗ 9.405
+ 42 ∗ 9.78) + 3,299 ∗ 9.90 = €2,022.3.
Note that this result is lower than the result without the adjustments ‘in
time,’ allowing for the fact that said adjustments would have added addi-
tional costs in terms of commissions, so it can be deduced that an approach
of this kind isn’t always the best; at least, not as shown here (with limited
adjustments).
The above procedure may have seemed somewhat convoluted and com-
plicated; but remember that, what the reader might have had most difficulty
with is calculating the capital of reference; and it’s unlikely this needs to be
calculated, as the value of the portfolio at the end of the day is shown on
the trading platform, along with the available liquidity, so all you have to
do is add the two figures to know how much money you should use in your
calculations. The above is used merely as an example.
The entire procedure could have been performed using the core equity
model or the reduced total equity model, but we won’t complicate things
with other calculations as the intention was just to show how you can con-
stantly check the volatility of the open position (or open positions).
An approach of this kind is also applicable, for all purposes, with
portfolios that follow the rules of a trading system.
228 ■ 11.4 Trading Futures

If you’re considering trading futures instead of shares, there are other

possibilities for applying money management techniques.
Using the fixed ratio method is very practical with futures as this method
doesn’t consider the parameters of the trade you’re thinking of opening but
just the value of your available capital compared to the initial capital.
The only difficulty is establishing the delta to use.
Initially, you could consider the stop-loss of the first trade and use a delta
that’s three times the stop-loss as long as it’s acceptable (a stop-loss of €100
would mean a delta of 300, which is clearly low, and you should always be
at least 2 percentage points from the initial capital).
Now let’s consider a discretionary trade with DAX futures and see how
to manage the trade on the basis of the chosen money management method.
Figure 11.3 shows the entry of the trade. On an uptrend, there was a
nasty reversal bar, a bar that opened with a gap up, reached a high of 5,249
then reversed its trend and took the market to new lows, reaching a low of
5,153 and closing very near to the same. A signal of this kind should start
FDXH6 LAST-Daily 02/07/2006 C=5689.0 +25.5 +0.45% O=5703.5 H=5715.5 L=5645.5 V=70360
5700.0
5600.0
5500.0
5400.0
BUY
5300.0
5249
5200.0
5153
5100.0
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.3 An example of a discretionary trade with DAX futures.

alarm bells ringing loudly for a probable reversal of the trend. The next day,
however, the market showed signs of uncertainty. The next bar, an up move-
ment, is in any case an inside bar (high smaller than previous high and low
greater than previous low). There is even a second inside bar again next day,
proving the considerable uncertainty of the market. At this point, we decide to
enter in the direction the market decides to take. We’ll enter long if the price
breaks the high of 5,249, or short if it breaks the low of 5,153, and these levels
would also be used as stop-loss levels when entering the trade. Immediately
next day there’s a bar with a strong up movement, and we enter the trade. At
this point, we only know the level at which we will be long (5,249) and the
level at which we’ll exit if the market changes direction (5,153).
In the previous period, the market didn’t show signs of a situation that
let us identify a target, so we decide to exit only in the case of a probable
reversal. We also decide to raise the exit point a little higher every now and
then to make the most of any profits.
This is what we’ll do a few days later, as shown in Figure 11.4.
Note that there’s a relative high, after which there are two down bars, the
second of which hits a low of 5,225.5 but then the market moves up again,
and as soon as the new relative high is broken we decide to raise the exit
level to 5,225.5.
230 The same principle will be applied in the future if there are other similar
cases, as in fact occurs as shown in Figure 11.5.

A new high is registered, and in the next bar the market reaches a relative
low of 5,276. The second bar after this one breaks the new high and we
decide to raise the exit level to the intermediate low of 5,276. At this point,
we’re in a situation in which the trade will make a profit in any case as we’ve
raised the exit level to be past the level we entered the trade at.
We continue as above, and soon there’s a new case, as shown in
Figure 11.6.
There’s a new high with a relative low at the very next bar. This low is
5,414. The market recovers, moving up again, and 2 bars later the low breaks
the old high. It’s time to raise the exit level to 5,414.
The same phenomenon occurs again, but this time the exit level that was
just set is reached, and the trade is closed, as shown in Figure 11.7.
Therefore, this is a new high, and with the next bar that has a relative low
of 5,509, the market recovers and two bars later breaks (by just a little) the
high recorded. So the exit level is raised to 5,509, a level that is broken when
the next bar drops.
5700.0
5600.0
5500.0
Passes new high 5400.0
5300.0
5225,5
5200.0
Raise exit level
5100.0
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.4 A broken relative high makes us raise the exit point to the intermediate relative low.
5700.0
5600.0
5500.0
5400.0
Passes new high
5300.0
5276
Raise exit level

5200.0
5100.0
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.5 Another breakout of the relative high with the exit point raised to the intermediate relative low.
5700.0
5600.0
Passes new high
5500.0
5414
5400.0
Raise exit level
5300.0
5200.0
5100.0
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.6 Raising the exit level again.

5700.0
Passes new high (just)
5600.0
SELL
5500.0
5509
5400.0
Raise exit level, immediately broken
5300.0
5200.0
5100.0
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.7 The new exit level is broken immediately.

Note that after closing the trade, there are other down bars, but then the
market recovers.
The trade produces a profit with one contract of:
Gain = (5,509 − 5,249) ∗ 25 = €6,500.
Now let’s analyse how we could have applied money management tech-
niques to this trade.
Let’s assume we have the same available capital of €100,000 before open-
ing the trade.
First we’ll look at the fixed fractional method and ask ourselves how much
we want to risk on the trade we’re thinking of opening.
The entry would be at 5,249 with the stop-loss at 5,153, which is the
equivalent of 96 DAX points. Each DAX point is €25, so the risk in this case
is:
risk = 96 ∗ 25 = €2,400.
This means that, for every DAX contract we’re risking €2,400.
Now, to calculate how many contracts we can buy we have to decide what
risk percentage we want to adopt for the total capital.
In the previous case we opted for 2% (note that this figure is usually cho-
sen before deciding on the potential trade. It’s not linked to the trade but 235
rather the risk profile, which is set before you sit down in front of the com-
puter to look for ideas for trades).
So, let’s take the same trader as in the previous example, who doesn’t
want to risk more than 2% on each trade. In Chapter 4, we introduced a
formula that can be used to calculate the number of contracts for the fixed
fractional method:
( )
(C ∗ f )
Maxloss
Therefore, in our case:
( )
(100, 000 ∗ 2∕100)
ff contracts = INT = 0!!!
2,400
What this means is the trade in question isn’t feasible, as the risk is greater
than the one we’re willing to take.
Remember that rejecting a trade for this sort of reason is part of money
management and shouldn’t be seen as something stupid. Money manage-
ment protects your capital from undesirable events and, in this case, losing
€2,400 instead of 2,000 is considered undesirable.
At this point there are four possible scenarios:
1. Ignore the trade and wait for a more suitable occasion for our risk profile.
2. Increase the risk percentage from 2% to 2.5% to ‘cover’ the potential
loss.
3. Increase the level of the stop-loss to a point at which we wouldn’t lose
more than €2,000.
4. Wait for a retracement to enter at a level at which we wouldn’t lose more
than €2,000 with the chosen chart stop.
Let’s consider the various options.
The first scenario is certainly the simplest, as you won’t have to think
about the trade anymore so you can calmly look at something else, but if the
market was favourable to the trade you were considering, this could lead to
regretting the decision taken, and for the next trades you might be tempted
to leave prudent money management aside.
The second choice, despite how logical and intelligent it may seem is,
in my opinion, the most imprudent course of action you could take. If you
decide to use a money management model you should always follow its indi-
236 cations. You can’t make changes once you’ve started just because you don’t
want to miss a trade. A 0.5% adjustment today could (in the case of a success-
ful trade) encourage you to make bigger adjustments tomorrow, meaning

all the good planning had been done in vain. This is definitely a choice I’d
strongly advise against.
The third option is a possibility, and if we enter the trade at 5,249 we
should set a stop at 5,169; in fact:
(5,249 − 5,169) ∗ 25 = €2,000.
But would this new level be in line with the considerations we made? Let’s
see in Figure 11.8.
The level would be just below the low of the second inside bar (its low
is 5,171). We could consider this level to be acceptable in chart terms for
this reason, but it would be pushing things for the reasons mentioned above.
The trend we entered the trade for would be invalidated when the oppo-
site end of the bar, with a low of 5,153, broke out; the fact that the new
level is just under the inside low is purely coincidental. What’s more, the
low of the first inside bar is lower than the potential new stop and could
act as a support in a fall. All the above would shake things up in terms of
FDXH6 LAST-Daily 02/08/2006 C=5646.5 -12.5 -0.22% O=5623.5 H=5656.5 L=5611.0 V=56405
5450.0
5400.0
5350.0
5300.0
5250.0
5200.0
5169 5150.0
5100.0
65.00
55.00
45.00
21 28 D 5
FIGURE 11.8 Hypothesis of a 5,169 stop-loss.

our discretionary decision undermining the basis on which the decision was
made, altering the concept of the trade. So, in my opinion, this isn’t a good
choice. There will certainly be some who might object, saying that if we take
option 1 we ignore this trade and look into others, and this solution could
be one of the other trades we find, but in the end there are many ways to
convince oneself a decision is right and this could also be one of them, and
in principle I wouldn’t recommend this route.
Last but not least, the fourth option. With a stop-loss of 5,153, in order to
set things up so that if this stop is triggered we don’t lose more than €2,000,
we have to use a maximum limit at which to enter the trade, this limit is:
limitebuy = (5,153 + 2, 000∕25) = 5,153 + 80 = 5,233.
In practice, we calculated how many DAX points are equal to €2,000,
dividing this amount by the value of each point (€25), then added this result
to the stop-loss exit level.
Those choosing this option would in effect place an order in the trading
book at 5,233 and wait for a market retracement to bring the price to this
level.
This is definitely the most advisable choice; although, there will be some
psychological repercussions. Entering on a retracement level that’s below
238 the set chart entry point isn’t easy because the chosen trend could appear to
be weakened. It can also be quite nerve-racking because there’s nothing to

guarantee the prices will reach that level and the wait could be in vain.
This is, however, the best solution in conceptual terms, and it’s the one
I’d recommend using, if possible.
What would happen in the case described? Let’s see in Figure 11.9. The
trade is entered a few days later.
At this point the fixed fractional method has no further effect and the
trade proceeds under its own steam with 1 DAX contract until the close,
which, as we saw above, is at 5,509.
In this case, the total result of the trade is:
Gain = (5,509 − 5,233) ∗ 25 = €6,900.
Even without the direct effect of money management (increasing the
number of contracts on the basis of the capital) it’s obvious we would have
made more than with the normal trade that made €6,500.
Now let’s take a look at what would have happened if we applied the per-
cent volatility model as the money management technique.
When we discussed the case of the MS trade, I emphasised how a per-
centage of 0.5% could be unsuitable for futures, at least with a capital of
€100,000.
Before we consider variations, let’s see what we would have obtained
using this percentage. First, let’s take a look at the ATR when the trade was
entered. Here’s the data in Figure 11.10.
The day before we entered, the trade the ATR was 53.60 points, equal
to:
53.60 ∗ 25 = 1,340.
With futures, you have to multiply the ATR by the value in euros of one
future’s point in order to obtain the volatility in euros. With the stock in our
list this wasn’t necessary simply because 1 point, in that case, was exactly €1.
Now, if we take this path with the percentage we considered at the time
of 0.5% we’d have a maximum limit range of:
range = 100,000 ∗ 0.5∕100 = €500.
Obviously, we’re already in difficulty limiting movements to €500 while
using an instrument in the portfolio with which we could move by as much
as €1,340.
240 If we wanted to trade in these conditions we could only do so if the ATR
was less than 20; in fact, with an ATR of 20 the volatility of the Future is
exactly €500 per contract (20 * 25 = 500).

This condition never arises in the period after the theoretical entry and
this goes to show that it would be fruitless to wait for volatility to drop before
entering. The chart also shows that it’s very rare to see an ATR of 20 with
DAX futures so, if you want to use it with this money management model,
you’d need a much bigger capital than the €100,000 we’re considering in
this case. We can say that the percent volatility model with a percentage of
0.5% is telling us DAX futures are not an option with a limited capital of
€100,000.
In any case, in order to study how the method would behave, we won’t
change the risk profile of the money management model, but will consider
we have an available capital of €500,000 and see how to make the trade.
As mentioned above, the volatility in euros of the future is €1,340 the day
before entry, so we calculate:
( )
500,000 ∗ 0.5
number of contracts = ENT = 1.
1,340 ∗ 100
5700.0
5600.0
5500.0
5400.0
5300.0
5200.0
5100.0
Average True Range(5) 86.60 100.00
90.00
80.00
70.00
53,60 62,9 56,5
52,8 60.00
49
50.00
39,9
30,3 40.00
21 28 D 5 12 19 26 06 9 16 23 30 F 6
FIGURE 11.10 ATR of 53.60 the day before entry.

TABLE 11.2
date ATR(5) Close of day Open next day
7/12/2005 62.9 5269.5 5233

14/12/05 56.5 5303.5 5300
21/12/05 49.0 5419 5418.5
29/12/05 30.3 5479.5 5467
5/1/2006 52.8 5543.5 5547
12/1/06 39.9 5544.6 5550.5
So we can buy just one contract.

If we don’t make any subsequent adjustments, this contract will be the
only one we have until closing the trade and nothing would change compared
to the initial example with one contract.
If, however, we want to apply what we did in the MS example and monitor
the trend of the volatility during the trade, we’d proceed as follows. First,
let’s see what data we need. The trade opens on 1 December 2005, then
we’ll monitor it every five trading days (Tables 11.2 and 11.3).
Let’s take things step by step:
242 The trade will in any case be closed on 13 January with two contracts, the
second one bought on market open, but in this case the trade is unfavourable.
The exit, as mentioned above, is at 5,509, with a total result of:
Gain = 505,212.5 + (5,509 − 5,249) ∗ 25 + (5,509 − 5,550.5) ∗ 25
− 500,000 = €10,675.
In this case, the diversified management of the contracts during the trade,
monitoring the volatility every week, produced decisively more profit!
To tell the truth, it must be stressed that adjusting the exposure on the
basis of the weekly volatility produced more profit than the case without
adjustments. We can’t, however, make a comparison with the fixed fractional
method shown above because in that case we were trading with an initial
capital of €100,000. If we want to make a comparison with the same initial
capital, we can calculate the number of contracts that could be used with the
fixed fractional method and a capital of €500,000:
( )
(500, 000 ∗ 2∕100)
ff contracts = INT = 4.
2,400
TABLE 11.3
7 Dec capital of reference 7dec = 500,000 + 1 * (5,269.5 - 5,249) * 25 = 500,512.5
= 62.90 * 25 = 1,572.5
volatility of Future in euros ( )
500.512, 5 ∗ 0.5
number of contracts = INT =1
1.572, 5 ∗ 100
14 Dec capital of reference 14dec = 500,000 + 1 * (5,303.5 - 5,249) * 25 = 501,362.5
volatility of Future in euros = 56.5 * 25 = 1,412.5

( )
501.362, 5 ∗ 0, 5
1.412, 5 ∗ 100
21 Dec capital of reference 21dec = 500,000 + 1 * (5,419 - 5,249) * 25 = 504,250
volatility of Future in euros = 49 * 25 = 1,225

( 504.250 ∗ 0, 5 )
1.225 ∗ 100
Now we can purchase a second contract when trading opens next day, in other words at
5,418.5!
29 Dec capital of reference 29dec = 500,000 + 1 * (5,479.5 - 5,249) * 25 +1 * (5,479.5 - 5,418.5) *
25 = 507,287.5
volatility of Future in euros = 30.3 * 25 = 757.5 243

( )
507.287, 5 ∗ 0, 5
757, 5 ∗ 100
A third contract is purchased when trading opens next morning at 5,467
5 Jan capital of reference 5jan = 500,000 + 1 * (5,543.5 - 5,249) * 25 + 1 * (5,543.5 - 5,418.5) *
25 + 1 * (5,543.5 - 5,467) * 25 = 512,400
volatility of Future in euros = 52.8 * 25 = 1320

( 512.400 ∗ 0, 5 )
1320 ∗ 100
We have to sell 2 contracts, which are sold when trading opens next day at 5,547.
Supposing they are the 2 purchased subsequently, this adds liquidity amounting to:
[(5,547 - 5,418.5) + (5,547 - 5,467)] * 25 = 5,212.5
12 Jan capital of reference 12jan = 505,212.5 + 1 * (5,544.5 - 5,249) * 25 = 512,600
= 39.9 * 25 = 997.5
volatility of Future in euros ( )
512.600 ∗ 0, 5
997, 5 ∗ 100
Another contract is purchased when trading opens next morning at 5,550.5
With four contracts, the entry would have been on level breakout and not
a retracement, and would have produced a final profit of:
Gain = 4 ∗ (5,509 − 5,249) ∗ 25 = €26,000.
As can be seen, we would have made a lot more profit in this case if we
used the fixed fractional method.
I mentioned above that with futures you can also use the fixed ratio
method, and it can also be a practical option.
As explained earlier in the book, the fixed ratio method starts with one
contract until sufficient profit has been accumulated to use two (this profit is
a quantity called delta and is set by the trader). In this specific case, therefore,
there are no limits in terms of initial capital, as you can start using the system
anyway. This could be a practical choice for those who don’t want to miss
out on any trades, but can be risky in the case of negative results (or worse
if there is a series of consecutive negative results).
In any case, there’s still the question of setting the delta, which, as we
don’t have any historical data available, is quite difficult to do. In this case,
the estimated loss was €2,400 (we calculated it in the case the stop-loss was
triggered), and as mentioned above, we could use this figure multiplied by
3, so €7,200, which we can round off to 7,000.
244 So the trade would be made in any case with one contract; as we’ve seen,
with one contract the trade produces a gain of €6,500, which means a fol-
lowing DAX trade would still have been made with one single contract, as
we haven’t increased the initial capital by the €7,000 necessary to start using
a second contract.
The fixed ratio method is particularly simple to apply because, in prac-
tice, it just considers the levels reached by the equity and doesn’t consider
the inherent risk of the trade you want to open, or the consequences of the
same on the trend of the equity while the trade is open. Obviously, the flip
side of the coin with management that’s simplified in this way is there’s no
real control of the entity of the risk, which, in certain cases, could be detri-
mental.
It’s plain to see that, for those who don’t want to get into complicated
calculations with futures, the fixed ratio method is definitely the way to go.
Those who want to have more control over the evolution of the trade, how-
ever, should adopt the fixed fractional method or even the percent volatility
model, periodically checking the volatility.
■ 11.5 Conclusions
Money management principles can also be applied to discretionary trading,
as long as you plan things well. Carefully planning the trade you want to open
isn’t only necessary in order to decide which money management method
to use; it’s also necessary to avoid unpleasant surprises or finding yourself
in situations in which there’s very little time to decide what to do about
something you missed.
The money management models we’ve discussed until now are applied in
the same conceptual way. The only difference lies in choosing the parame-
ters, which can’t be based on exhaustive historical data and must be chosen
by the trader in line with his or her feelings and disposition to risk.
If you use the percent volatility model, while the trade is open you can
monitor the level of exposure on the basis of the current volatility and, by
using suitable calculations, adjust the number of contracts used. This pro-
cedure can involve some difficulties but also be very effective to make the
most of market trends. This type of method can, of course, be used both for
automatic systems and a series of trades chosen personally by the trader.
245
CHAPTER 12
Questions
and Answers
Does money management study the trades to place, and provide for the strict applica-
tion of a stop-loss and a set profit target?
246 No, money management only tells you how much to invest. It doesn’t tell
you where and how to exit a trade. Trade management and studying trades
is part of ‘risk management.’
But in some cases money management tells me not to open a trade. Isn’t this the same
as specifying entry and exit levels?
No! If the chosen money management model advises you against opening
a certain trade, it does so because it finds the risk of that trade excessive for
the risk you are willing to take. In this case ‘how much’ is equal to 0.
The strategies in the book produce astonishing results. Can they be used?
The strategies in the book are basic models used to show how to apply
the rules of money management. They are trend-following strategies that
worked in the past and might work again in the future; however, I wouldn’t
recommend them to anyone as they’re very basic, and there could be long
periods in which they don’t produce profits. It would be better to concen-
trates one’s efforts on developing something more solid.
Can you really obtain these kinds of results using money management?
Yes you can, the correct application of money management techniques
produces notable results; but everything takes time, and you shouldn’t
expect to emulate Larry Williams and increase your capital a hundredfold
in just one year. Throughout the book, medium to long-term investments
have been taken into consideration, and in these periods there were highs
and lows. It can be hard to get through the bad times without changing
everything and starting from scratch again.
When you refer to the ‘risk’ of a position, do you mean the capital invested in that
position?
No, the risk is how much you could lose on the current trade, unless
you’re considering losing the whole investment (this could occur if you’re
working with options), the risk is just the part that would be lost if a stop is
triggered.
A strategy with a 5% risk is called aggressive. 5% doesn’t seem like much to me, so
why is it called aggressive?
5% could be considered a medium–low figure; but if this is the percent-
age you’re risking on every trade, a sequence of bad trades could produce
a drawdown that would be difficult for most traders to withstand. A lower 247
risk percentage certainly makes it easier to get through bad periods without
QUESTIONS AND ANSWERS

losing sleep.
I heard that Ryan Jones, who invented the fixed ratio method, entered the Robbins
contest and ended up with a 95% drawdown. Is this true?
The results of the World Cup Trading Championship can be sur-
prising – surprisingly good or surprisingly bad. A money management
technique can be considered extremely important but it won’t stop a
trading strategy going bad. The Fixed Ratio money management method,
for example, doesn’t stop a strategy due to insufficient funds but continues
with one contract for as long as there’s capital to trade with. If you’re
competing in a championship, you need to use an aggressive approach,
whatever the technique you’re using. The Fixed Ratio method helps limit
drawdowns when you’ve accumulated capital but it’s more aggressive at the
start, and this can produce extremely negative results.
Can you write the codes directly in EasyLanguage with the formulas to calculate the
number of contracts, so you don’t have to do so before each trade?
If you want to do so you can, but I wouldn’t recommend it. If you’re
back-testing you could, but in real time there are limits. It’s nice to have a
signal from TradeStation with the number of contracts to use; but you need
to remember that TradeStation considers you’re following the strategy to
the letter. Its calculations are based on the capital that’s grown with all the
registered trades. If, for any reason (an error, a holiday, or bad mood) you
miss a trade, your available capital won’t be the same as that in the list used
for the calculations, and this could change the results considerably.
If I started with a strategy and things are going well but I have to withdraw funds
from the capital to cover personal costs, what should I do?
This isn’t a problem. If you have to withdraw funds you simply adjust
the available capital. If you’re using a fixed fractional method you do your
calculations for the next trade with the remaining capital. The same goes for
the percent volatility and fixed ratio methods. In other words, it’s as if you
lost the amount of money you’ve withdrawn in a losing trade.
248 If I find myself with additional liquidity and add it to the capital I’m trading with,
do I have to start from scratch?
No, the same reasoning applies as in the previous question. A new source
of capital can be considered in the same way as a winning trade and you
continue to trade with the new capital as if you made the profit trading.
Just one thing to watch out for, both in this case and the above. If you’re
using filters for the trades based on the position of the equity line in relation
to its average, you should adjust the calculations as if the amount added (or
withdrawn) was added or subtracted from the start, so this amount doesn’t
affect the 30-day average, or that of any other period.
In the book you say someone recommended using the Kelly formula and using 50% of
the resulting figure to calculate how much to invest. Isn’t it simpler to do just that?
Often, even 50% of the Kelly percentage can be too aggressive, depending
on the system to which the method is applied. Kelly referred to systems that
weren’t used for trading. The application of this formula on the stock market
is something of a strained interpretation, and it’s much better to use other,
more suitable, methods.
But Larry Williams made sparks fly with the Kelly formula, so if it worked for him,
why shouldn’t I use it, too?
Larry Williams is a brilliant trader, but the year he won the championship
with over 10,000% was a particularly favourable year for him. Let’s just
say that, as well as using excellent techniques and the Kelly formula, Larry
Williams also had a good deal of luck on his side when he reached that result.
You must remember that, in the same year, he passed the $2 million mark
and then his funds quickly fell to 700,000 USD, which was the only bad
period that year. If this had happened in the first month, he wouldn’t have
finished the championship. In the following years, the Kelly formula revealed
its true nature and Larry Williams stopped using it in favour of the fixed
fractional method, taking the maximum historical loss of the system as the
loss of reference.
I like the fixed ratio method. Why does no one talk about it in the trading world?
The fixed ratio method isn’t used much as a technique in the professional
trading world because, at first when using the strategy it can be too aggres-
sive. Many professional traders prefer to take things easy at the start; but it
must be said that they usually have more than enough capital from the start,
so have no qualms about starting with a fixed fractional method set up in a
249
conservative way.

The fact that the number of contracts is obtained from the square root
of the capital is considered a sort of mathematical ‘alchemy’ that makes this
method somewhat more mysterious, that’s all. In reality, as in all things, it
has its advantages and if you like it there’s no reason not to use it.
You mention the drawdown a lot, but how important is it really? If I can increase my
capital tenfold what do I care if it’s a rollercoaster ride to get there?
Whenever one mentions increasing their capital tenfold or any other
incredible performance, they’re always talking about the past. If a system is
good, it will probably also be good in the future, but no one can guarantee
that for sure. If we knew that after a certain period of time the capital would
increase by a certain amount, of course we wouldn’t worry how much it
fluctuated to get there, but no one knows what will happen and the final
result is just hypothetical even with what we might consider to be a high
probability of success. Too big a drawdown will inevitably result in a period
of sufferance, raising doubts about the validity of the system you’re using to
make trades.
I want to stop taking trades from the strategy if the equity drops below its average. Do
you always have to use a 30-period average?
The 30-period average is just one of the many solutions adopted – 30
periods provides a good response and considers a sufficient period of trades.
A lot depends on the number of trades the strategy usually makes in a year.
If, to take an absurd example, the strategy shows less than 30 trades/year, it
would be better to use a lower period. In any case, this type of stop is chiefly
used as a precautionary measure against the ‘death’ of the system. It’s not
really used to improve performance, and other methods could be used such
as the breakout of a low after 30 (or another number) of periods.
If I really want to use 30 periods, should I consider the theoretical trades of the strategy
or those I’ve actually made. I missed a few. Should I consider them or not?
As the stop is used to block the basic strategy and not specific trading,
all the calculations should be made on the basis of theoretical trades, the
ones in the report in other words. These are the result of the unconditional
application of the rules. If you miss some trades, that’s another problem and
nothing to do with the strategy.
Why in the book do you say you don’t really believe much in trade dependency and
250 using the z-score?
These studies enter into a realm that’s overly ‘statistical.’ If you apply
mathematics to trading too much, you run the risk of turning it into some
sort of scientific monster that loses its foundations. It’s always better to keep
things simple, and there are already more than enough possible complica-
tions with standard money management solutions, so including these kinds
of studies in my opinion is excessive.
I’ve got a great strategy with an 80% success rate thanks to some special filters. Can
I use the Monte Carlo simulation to prepare a good plan of action?
It’s certainly possible but, in general, if you’re using a heavily filtered strat-
egy, it’s a good idea to study the trend without the filters and apply the Monte
Carlo simulation to the results of this approach. The results will definitely
be worse, but closer to the real situation. In principle, having cleaned the
strategy of losing trades with filters that make sense, you run the risk of dis-
torting the studies by making the results too good. Let’s say the filters should
be applied later, to improve a strategy that’s already basically a good one.
In the work plan is it more important to plan profits or drawdown limits?
Profits should definitely be an objective, but the range of potential vari-
ations of the same is often very wide. Let’s say that, if after the period
considered the profits are below the limit of two standard deviations from
the mean, then the strategy isn’t working very well. The same could be said
though if profits were over the limit of two standard deviations from the
mean. Such a good performance should start alarm bells ringing, because it
could mean you’re trading too aggressively on the market.
Drawdowns, on the other hand, are more important, and the range of the
variation is narrower. Therefore, it’s easier to see whether the point you’re
at is within acceptable limits.
If I want to use a portfolio of stocks and futures, is it best to use the same strategy for
everything?
The best solution would be to use different strategies that diversify not
only the instrument but also the logic. For example, with some instruments
you could trade using trend following strategies such as the channel breakout
or (as often mentioned in the book) a moving average crossover. With other
instruments, you could adopt strategies in which trading is based on a price
pattern or study countertrend strategies. In this way, you’ll have the greatest
diversification and optimal synergy with the instruments. 251

Is it very important to limit global exposure?
Yes, it is, and it’s a good idea to dedicate the time required to it. In the
same way as you limit risk for a single position, it’s just as important to limit
total risk. If you open too many positions you’ll go from risk control to a
strategy of diversification but lose the importance of the exposure. Limit-
ing the number of positions, on the other hand, maintains the initial money
management basis intact to avoid overexposure on the market.
Can trading give you a constant monthly income?

Alas, you’ll often see the promise of a monthly income in adverts the aim
of which is to sell you a certain type of service or a certain product. Profits
made from trading don’t guarantee such a level of stability for them to be
comparable to a monthly wage. There could be months, or even years, that
close at a loss and you need to cover your back to be prepared for such things.
APPENDIX I
■ I.1 The Impact of a Trading System

on Planning
The classic approach described in this book involves applying a money man-
agement model to a system, or portfolio of systems or instruments, to max-
imize their effectiveness.
252 As mentioned in the book, money management won’t transform a losing
system into a winning one, or eliminate particularly critical points inherent
to the same system; however, creating a winning model after developing the
basic system may not always be the most effective way to exploit the profit
amplification effect.
If we know the intervention dynamics of the money management algo-
rithm, it may be possible to reconsider some aspects that marked the stages
of development of the operating rules, considering not the single trading
system as the final product but rather, the whole system + position sizing
package.
In order to better understand this aspect, let’s take a specific example
to see how a change made to the classic approach can produce indubitable
benefits.
■ I.2 The Trading System
For the first stage, we’ll develop a banal automatic trading system and take
a look at the considerations that resulted in the final product.
For our example, we’ll use Euro Stoxx 50 futures traded on EUREX. This
market is one of the most liquid in the world, so it’s a good example of an
underlying that can be traded with a growing number of contracts.
Every tick of the instrument is equal to one point of the same, with a
value of 10 euros.
We’ll create a system to be used only for intraday trading based on Toby
Crabel’s well-known Opening Range Breakout principle. This principle
attempts to follow a trend that sets in during the day and is based on
the principle of entering a trade at the market price when it has already
moved in one direction by a certain amount, generally greater than the last
movements.
In general, this approach involves studying daily bar charts, and entry
points are defined in various ways, by measuring a movement from the open
of the day, hence the name Opening Range Breakout, which refers to the
breaking of a range starting from the open value. These entries though may
need monitoring for more than one day and this isn’t what we’re aiming for
253
in this example; so, in line with the proposed intraday logic, we’ll take a
APPENDIX I
detailed look at each single day in order to study the entries. To do this,
we’ll use a 60-minute chart.
Using the intraday bar chart, we can check whether it’s a good idea to
trade only within a certain period of time rather than generically trading any
time the market is open; to do this, in the code, we enter the ‘StartTime’ and
‘EndTime’ as inputs for the time limits within which you want to consider
any breakouts.
The breakout levels can be defined on the basis of the previous day’s range,
multiplying it by a value we can arbitrarily call ‘MyPerc,’ which will be sub-
ject to study, as well as the above-mentioned times.
The EasyLanguage code follows:
Input: MyPerc(0.5, MyStopLoss(0), StartTime(0), EndTime(2200);

var: MaxSetup(0), minSetup(9999), MyRange(0), Mycontracts(1);
if date<> date[1] then begin
MyRange = highd(1)-lowd(1);
MaxSetup = opend(0) + MyRange*MyPerc;
minSetup = opend(0) – MyRange*MyPerc
End;
if Time > StartTime and Time < EndTime then begin
if close < MaxSetup then buy MyContracts contracts next bar at MaxSetup stop;
if close > minSetup then sellshort MyContracts contracts next bar at minSetup stop;
end;
setstopcontract;
if MyStopLoss > 0 then setstoploss(MyStopLoss);
setexitonclose;
FIGURE AI.1 Initial strategy EasyLanguage code.
The above code buys over a certain level, calculated by adding a portion of
254 the previous day’s range to the opening price; or sells if there’s a loss equal
to the same portion of the previous day’s range in relation to the opening
APPENDIX I
price.
As entries can be triggered only after a certain time (further on we’ll
study the optimal time, called StartTime), there’s an additional condition
tied to the fact we still need to be below the buy long entry level (close <
MaxSetup) or above the sell short entry level (close > MinSetup) in order
to take the trades into consideration (this prevents entering the market after
a breakout that occurred outside the set times).
The system closes the positions at the end of the day (setexitonclose) and
has a monetary stop for each contract that’s only used if the relevant input
(MyStopLoss) is other than 0. (Note that currently the system is tested with
one single contract as each Money Management algorithm will only be con-
sidered subsequently.)
Until December 2018, the market in question had been trading continu-
ously from 8:00 to 22:00. Yet before June 2006 it traded at different times,
with the market opening at 9.00 and closing at 20.00. Extended times were
first adopted on 1 June 2006; therefore, we’ll focus our studies starting from
that date to simplify the analysis. These studies also include 15 euros in trad-
ing costs per side (the hypothesis is for one tick of slippage and a commission
of 5 euros per contract).
For the first study, let’s make a hypothesis of working for the whole day, so
with a StartTime = 800 and an EndTime = 2200, checking which portion of
the previous day’s range produces the best results. In Figure AI.2 we can see
that the entry points that are 25% higher than yesterday’s range are those
that produce the most profit, so we’ll use a MyPerc value = 1.25 for the
subsequent analyses.
Note that the chosen value does represent the absolute optimal value, but
the net profit chart in Figure AI.2 shows we’re not dealing with an unsta-
ble value as the results on either side (1.2 and 1.3) are in any case similar.
Therefore, this is a case in which choosing the best value deriving from the
optimisation isn’t necessarily a risk.
Now, with a joint optimisation of StartTime and EndTime, changing these
values every hour, we can see how the best results can only be obtained with
a start time after 10.00 and an end time before 18.00. The results are shown
in order of decreasing net profit, and Figure AI.3 shows the most interesting
results.
15000
10000 255
APPENDIX I
5000
0
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
0.55 0.65 0.75 0.85 0.95 1.05 1.15 1.25 1.35 1.45 1.55 1.65 1.75 1.85 1.95
–5000
–10000
–15000
–20000
FIGURE AI.2 Net profit trend as MyPerc (which is multiplied by yesterday’s range)
changes.
k
StartTime: EndTime:
Input Input
FIGURE AI.3 Profit trend on the basis of the chosen trading time windows.
Input: MyLongDay(0), MyShortDay(0);
End;
if close < MaxSetup and dayofweek(date) <> MyLongDay then buy MyContracts contracts next
bar at MaxSetup stop;
if close > minSetup and dayofweek(date)<>MyShortDay then sellshort MyContracts contracts next
bar at minSetup stop;
end;
setstopcontract;
setexitonclose;
FIGURE AI.4 New EasyLanguage code excluding some days of the week
The work we’re doing is entirely closed by the end of the day with all the
open positions closed. Therefore, it may be a good idea to conduct another
study to find out whether there are any days in the week that aren’t very 257
well suited to long positions and others that aren’t good for short positions.
APPENDIX I
Figure AI.4 shows the EasyLanguage code changes.
The optimisation of each datum from 0 (trading every day) to five shows
whether it’s advisable to exclude some days. The results are shown in
Figure AI.5.
Figure AI.5, in fact, shows how, as regards situation ‘0.0,’ there are var-
ious combinations that give better results. As we’re performing more of a
clean-up operation rather than looking for the most consistent profits, it’s
important to make a choice on the basis of the quality of the remaining
trades. The results are therefore ordered starting from those with a higher
average trade and excluding Friday long trades and Wednesday short trades,
which improves not only net profit (from 16,000 to 16,580) but, above all,
the average trade (from 68.67 to 85.91).
The last thing to do at this point is to check whether using a stop-loss will
produce further improvements.
An optimisation from 0 to 1,000 (0 equal to no stop-loss) produces the
results in Figure AI.6.
Figure AI.6 shows the best result is obtained without a stop-loss, and it’s
also evident that very tight stop values produce quite disappointing results.
MyLong- MyShort-
Day: Day:
Input Input
FIGURE AI.5 Excluding Friday from long trades and Wednesday from short trades improves the results in average trade terms,
and also net profit.
StopLoss:
Input
FIGURE AI.6 Trend in profits as the stop-loss changes.

The above shouldn’t give us the wrong idea though that it’s better to trade
without a stop-loss. In fact, the above strategy has a time stop and positions
are closed by the end of the day. Therefore, there are no cases in which losing
positions remain open in the hope they’ll recover. This, in fact, is the cause
of more than a few disasters, and almost always locks up trading. There must
always be a stop, in other words an action that closes both winning and losing
trades. It may be a monetary stop (which in this case proves less effective), a
time stop (as used in this example) or a condition of stop and reverse (typ-
ically found in some trend following strategies always in the market). Hope
should never be a reason for keeping a position open.
The basic strategy can be considered to be completed, and Figures AI.7
and AI.8 respectively show the performance report and the equity line.
This certainly isn’t a strategy that’ll make us millionaires, but there’s no
doubt that, even using an extremely simple approach, we were able to make
a profit.
As mentioned above though, the chosen market is well suited to trading
also with a significant number of contracts, so we’ll proceed by applying one
260
APPENDIX I
FIGURE AI.7 Performance report of the strategy developed.

Equity Curve Line - @FESX 60 min.(06/01/06 09:00 - 01/18/13 22:00)
18000
16000
14000
12000
10000
Equity(€)
8000
6000
4000
2000
–2000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
FIGURE AI.8 Equity line of the strategy developed.
of the money management models described in the book, using the classic 261
percent f, and assuming we start with a capital of 100,000 euros (with the
APPENDIX I
available data we could decide to use much less capital with this threshold,
but in this appendix the idea isn’t to verify the trading limits of the strategy
we’ve just developed).
Entering the set of instructions necessary to calculate exposure in the
EasyLanguage code shown in Figure AI.9, we add the data on risk and cap-
ital as well as the ‘MaxLosing’ input, which is simply the max. loss ever
recorded. In our case, as can be seen in Figure AI.7, this value is 630 euros.
Applying this set of rules to the above system changes the results, as shown
in Figures AI.10 and AI.11.
On the basis of whether the scenarios in Figures AI.10 and AI.11 are
acceptable, we’ll have finished the ‘classic’ development.
However, if we consider how contracts are calculated using the percent
f method, a lot depends on the entity of the maximum loss. So, returning
to the table in Figure AI.6 we’ll see that, by adopting a monetary stop value
of 350 euros (35 ticks of the instrument), it’s true that net profit drops by
about 23% (12,730 euros compared to 16,580 euros) but the max. loss is
380 euros (350 stop-loss + 30 trade costs), which could open interesting
opportunities when calculating contracts using percent f.
Input: MyLongDay(0), MyShortDay(0);
Input: MyPercRisk(2), MyStartCapital(100000), MaxLosing(630);
var: MyEquity(0);

MyEquity = MyStartCapital + netprofit;
If MaxLosing >0 then MyContracts = intportion(MyEquity*MyParcRisk/100)/MaxLosing);
End;
if close < MaxSetup and dayofweek(date) <> MyLongDay then buy MyContracts contracts next
bar at MaxSetup stop;
if close > minSetup and dayofweek(date)<>MyShortDay then sellshort MyContracts contracts next
bar at minSetup stop;
end;
setstopcontract;
setexitonclose;
262 FIGURE AI.9 New EasyLanguage code with the instructions for calculating the
contracts with percent f.
APPENDIX I
Therefore, in the hypothesis of using a fixed stop of 350 euros per contract
we’ll recalculate the profits deriving from the application of the percent f
model with a 2% risk (this time the MaxLosing parameter will be set at 380).
Figure AI.12 shows the performance report and Figure AI.13 the equity line.
It’s immediately obvious that, in this case, the version with the stop-loss
produces much better results than that with a simple time stop. The low-
est ever loss, in fact, permits a greater exposure for the same risk, and the
impact on the final results is all but negligible.
Trading with a 2% risk may be reasonable but, in order to highlight the
above phenomenon let’s conduct some tests also with a risk of 5% per trade.
In Figure AI.14 the performance report has no stop-loss and therefore the
max. system loss is 630 euros per contract. Figure AI.15 shows the relevant
equity line.
Now let’s perform the same operation with a 350 euros stop-loss per
contract. Figure AI.16 shows the performance report, and Figure AI.17 the
equity line.
FIGURE AI.10 Results with percent f equal to 2%. 263
APPENDIX I
70000
60000
50000
40000
Equity(€)
30000
20000
10000
–10000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
FIGURE AI.11 Equity line with percent f equal to 2%.

264 FIGURE AI.12 Performance report with a 350 euros stop-loss per contract and a
percent f model set at 2%.
APPENDIX I
90000
80000
70000
60000
50000
Equity(€)
40000
30000
20000
10000
–10000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
FIGURE AI.13 Equity line with a 350 euros stop-loss per contract and a percent f
model set at 2%.
FIGURE AI.14 Performance report with a percent f model set at 5% and a simple 265
time stop.
APPENDIX I
260000
240000
220000
200000
180000
160000
140000
Equity(€)
120000
100000
80000
60000
40000
20000
–20000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
FIGURE AI.15 Equity line with a percent f model set at 5% and a simple time stop.
266 FIGURE AI.16 Performance report with a 350 euros stop-loss per contract and a
percent f model set at 5%.
APPENDIX I
400000
300000
200000
Equity(€)
100000
–100000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
FIGURE AI.17 Equity line with a 350 euros stop-loss per contract and a percent f
model set at 5%.
The differences are obvious and, even though the trend of the equity line
isn’t very stable, it’s evident that greater exposure for the same risk can pro-
duce a notable increase in final profit.
The above derives from considerations based on the experience of the
individual developer. The choice of one money management model over
another, whether you need to change the basic settings of the strategy,
remains entirely in the operator’s hands. In fact, it’s very hard to suggest
a correct development path that’s valid for any approach, and only
deep-rooted knowledge of money management dynamics will give you the
right intuition for one particular strategy being developed.
267
APPENDIX I
APPENDIX II
■ II.1 Understanding the Type of Strategy

The money management approach linked to a portfolio of systems has shown
how different models can produce significantly different results. The simple
possibility of adopting a different model for the portfolio of systems each
time, has also been considered.
268 While this was logical in the case of a ‘common engine’ system applied to
several instruments, it can be a weaker approach if systems with significantly
different features are combined.
As usual, a specific example is the best way to illustrate the possible
options to handle situations of this kind in a more effective way.
In the previous appendix we traded Eurostox50 futures using an intraday
strategy. The conclusions we reached show that, the combination of strategy
and position sizing model may even raise doubts about the validity of the
basic strategy structure. In the end, we also found that a strategy with a
monetary stop-loss was preferable as, although less effective than a version
with just a time stop, it does prove to be more effective once the percent f
model is applied.
We’ll use this version of the strategy in this appendix, in the hypothesis
of using it with another, quite different, strategy.
Despite being applied to Eurostox50 futures, the new strategy won’t be
based on 60-minute bar charts but rather 240-minute charts, and will simply
open countertrend trades when a min. or max. period is reached.
The basic code is that shown in Figure AII.1.
If marketposition <= 0 then Buy mycontracts contracts next bar at lowest(low, 15) limit;
If marketposition >= 0 then sellshort mycontracts contracts next bar at highest(high, 25) limit;
FIGURE AII.1 Countertrend strategy code.
This strategy, in the period that goes from 1 June 2006 to January 2013,
produced approximately 40,000 euros in 266 trades (also in this case 15
euros in trading costs were considered). Figure AII.2 shows the performance
report.
Now, in the hypothesis of combining the two systems and using them with
an initial capital of 100,000 euros, it’s immediately obvious that a percent f
based on the max. system loss won’t produce great benefits for the combined
forces, as the max. loss of the new system is 5,240 euros, which requires a
notable amount of available capital or a particularly high percentage of risk
per trade.
269
APPENDIX II
FIGURE AII.2 Countertrend strategy performance report.

In order to make a comparison with values considered ‘reasonable’ until
now, we’ll analyze the results with a 2% risk.
On the basis of what we’ve learned so far, we can see that in order to use
the new system with a 2% risk per trade, we need a capital of at least:
( 5,240 )
∗ 100 = 262,000 euros.
2
Therefore, with 100,000 euros we can only start to trade using the second
system if the equity produced by adding the profits of the first one is over
the threshold, we’ve just calculated.
Figure AII.3 shows the portfolio equity line chart and the relevant draw-
down data.
As can be seen from the trend of the line, the level required to start trading
with the second system is never reached. In fact, the result obtained doesn’t
allow for the combination of the two systems at all, and is a mere exercise in
applying a position sizing model to the trading system shown in the previous
appendix.
Increasing the risk produces two effects. The first is a faster and more
noticeable increase in terms of capital from the first trading system, while
the second is the need for a lower level of liquidity in the second system to
270 start trading.
With a percent f model set at 2.5%, for example, this level is equal to:
APPENDIX II
( 5,240 )
∗ 100 = 209,600 euros
2
175,000
Equity ($)
150,000
125,000
100,000
Drawdown (%) Drawdown ($)
–10,000
–20,000
–5
–10
Date
Equity High ($) Equity Low ($) Equity Drawdown ($) Equity Drawdown (%)
FIGURE AII.3 Portfolio equity line of the two systems with a percent f model
set at 2%.
200,000
Equity ($)
150,000
100,000
Drawdown ($)
–20,000
Drawdown (%)
–10
Date
FIGURE AII.4 Portfolio of the two systems with a percent f model set at 2.5%.
Figure AII.4 shows how said level is (just) exceeded several times.
However, if we measure the contribution of the two systems to the line
in Figure AII.4 we’ll see that of the 107,840 euros in profit (the line ends
at 207,840 euros), only 2,720 were produced by the countertrend trading
system, and in particular in 20 trades. Therefore, its contribution is still 271
negligible, and it’s difficult to consider this as a portfolio also in this case.
APPENDIX II
It’s only with a risk percentage of 5.24% that the second system can trade
from the start (5.24% of 100,000 covers the entire max. loss of the sys-
tem, equal to 5,240 euros). In reality, we need to check which system starts
trading first and, if it’s the first, the level the equity line reached when the
second could start trading, in order to evaluate the exact percentage (the
initial capital of 100,000 euros may have increased or decreased in the mean-
time); but even this isn’t enough as, while trading with the two systems,
there may have been losses that reduced the available capital to below the
100,000 euros mark. Therefore, we’ll take the 5.24% risk as a purely math-
ematical exercise and evaluate the results of the portfolio of systems shown
in Figure AII.5.
Figure AII.6 shows the equity line, and in Table II.1 we see the contribu-
tion of each system.
As can be seen, the second system managed to make all the 266 trades
planned and its contribution was just over 20% of total profits.
FIGURE AII.5 Result of the portfolio of the two systems with a percent f model set at
5.24%.
600,000
Equity ($)
400,000
200,000
272
Drawdown ($)
APPENDIX II
–100,000
Drawdown (%)
–20
Date
FIGURE AII.6 Equity line of the portfolio of the two systems with a percent f model
set at 5.24%.
TABLE AII.1 Results of the two systems with a percent f model set at 5.24%.
Net Profit Number of Trades
System 1 399.85 196

System 2 122.64 266
FIGURE AII.7 Results of the portfolio with a 7% ‘daily’ risk.
If we want to compare the effect of each single system on the portfolio, we

could consider the concept of max. loss in a different way and, in particular,
not in terms of a single trade but a period of time.
For example, we might consider how much, in the available history, is the
max. daily loss of each single system. The ‘daily’ reports can be found in the
reports of each single system, and in these reports it’s easy to see which was
the worst day for the system.
The ‘weight’ of a single day makes sense, particularly if your trading 273
results are measured on a daily basis. In this case, it’s very important to
APPENDIX II
parametrize everything for the effect of each single day. If weekly or monthly
measurements are made, we can take these time slots as a sampling period.
Analyzing the performance report, we can see that system 1 had its worst
day with a loss of –700 euros, while system 2 had its worst day with a loss
of –3,310 euros. (These values cannot be deduced from the data reported
in the book figures shown so far, and were obtained directly by the author
from detailed trading system reports.)
In order to know whether the data will be useful to create a somewhat bet-
ter scenario, it’s necessary to push the risk per single day toward values that
are able to produce similar results to the best case yet in terms of net profit.
In particular, trading with a 7% risk produces the results in Figures AII.7
and AII.8.
A comparison of the percentage drawdown values of the two approaches
clearly shows the improvement obtained when taking the worst day as the
parameter of reference as far as risk is concerned. Also, a simple glance at
the two equity lines confirms the better trend of the one in Figure AII.8.
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–50,000
Drawdown (%)
–10
Date
FIGURE AII.8 Equity line of the portfolio with a 7% ‘daily’ risk.
In both cases, the levels used for the risk percentages are obviously merely
indicative and only useful for the purpose of this mathematical exercise.
Trading with levels over 5% for each single loss or 7% for each single day
is, without a doubt, an ill-advised approach. Although the drawdown levels
274 shown in this analysis might be considered tolerable by many readers, risk
levels of this type represent a real hazard for potential future scenarios.
APPENDIX II
It becomes more logical to accept higher risk percentages if we analyze

the max loss over a longer period of time. Therefore, we’ll run a monthly
analysis, and for strategy 1 this will show a max. loss of 880 euros, while for
strategy 2 the max. monthly loss is 5,500 euros. (N.B.: The max. losses can
be found in the detailed trading system reports of each single contract.)
Therefore, applying a max. monthly loss limit produces results similar
to those in Figures AII.5 and AII.7. Trading with a 10% risk percentage, in
Figures AII.9 and AII.10, produces the results shown.
Once again, the results based on time slots show more acceptable draw-
down levels, and a more regular equity line trend than those based on single
trade max. system losses.
In this case, there would be a scenario that could also be used, as a 10%
per month risk might be considered acceptable, and in any case, it’s definitely
more logical than 7% per day or 5.24% per single trade.
Lastly, we should also give due consideration to the type of strategies. The
first is an intraday strategy and it could be a good idea to measure the monthly
FIGURE AII.9 Performance report of the portfolio with a 10% ‘monthly’ risk.
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–50,000
275
–100,000
APPENDIX II
Drawdown (%)
–10
–20
Date
FIGURE AII.10 Equity line of the portfolio with a 10% ‘monthly’ risk.
performance, as this ‘averages’ out the daily results over a more reliable
period of time. The second strategy however is always in the market; the
positions are simply reversed with opposing entry points. Risk on a monthly
basis can certainly make sense, but it has less impact compared to what each
individual position could consistently have on the equity line. As mentioned
in the previous chapters, one way to measure the effect of system positions
on equity is to size exposure using the percent volatility method.
Therefore, we’ll consider the case in which the first strategy is weighed on
a monthly basis (as in the previous example) with a 10% risk (which we found
reasonable), while the second strategy is applied counting the contracts in
FIGURE AII.11 Results of a mixed approach based on types of strategy.
600,000
Equity ($)
400,000
200,000
Drawdown ($)
–50,000
276
–100,000
APPENDIX II
Drawdown (%)
–10
–20
Date
FIGURE AII.12 Equity line with a mixed approach.
order to prevent the equity line fluctuating by more than 0.125% per day
(the average range of potential fluctuation was calculated as the average true
range for the last five days), and 0.125% was chosen as it’s the value that
produces results similar to those in the previous examples.
Figures AII.11 and AII.12 show the results of this approach.
As can be seen, for the same drawdown, the latter approach produces
slightly better results, which confirms its validity.
If we wish to further refine this, we could use a 9% monthly risk for the
first strategy and a percent volatility of 0.15% per day for the second on the
average true range in five days. This approach would bring net profit to just
over 600,000 euros with an increase in max. drawdown with the open posi-
tion of around 25%. This however is an exercise for optimising past trades,
which doesn’t leave a great deal of opportunity for similar improvement in
real-time future trades.
What we wished to illustrate in this appendix is how important it is to
consider also a model that produces the best possible contribution to the
basket from each single strategy. The example shown is limited to the com-
bination of two strategies only to show the different results. If we apply this
logic to a more numerous group of systems, there would be concrete benefit
in terms of final results.
277
APPENDIX II
APPENDIX III
■ III.1 The Advantages of Forex

Today the online trading market is literally teeming with all kinds of ads
promoting Forex trading.
Unfortunately, most of these ads give people the idea that there’s some
sort of ‘wonderland’ where you can get rich quick, often after simply reading
278 a few explanatory notes on how the market in question works. In many years
of trading, one thing I’ve learned is that nothing comes easy in trading, and
if people approach this profession with the idea of getting rich their results
will often be a letdown, if not totally disastrous.
Often in this book I’ve emphasised the importance of limiting losses and
conserving capital, and profits will be an almost natural consequence of a
job well done.
But I don’t want to demonize Forex because, if used with due caution, it
can also offer some considerable advantages.
Strictly in terms of trading, it’s undeniable that the variety of the offer on
the Forex market means notable diversification, not only in terms of instru-
ments but also in terms of actual approaches. Various currency pairs are
well-suited to trend trading, while others tend to revert to the mean value,
which makes opening countertrend positions preferable. It’s important to
note that while trend positions could in theory have an infinite duration (once
a trend has started it could, at least in theory, last forever), it doesn’t make
sense to consider countertrend positions the same way, because if a position
opened in countertrend was intended to last for too long, it would by def-
inition become a trend position, as the market would be moving in a new
direction.
The concept of countertrend in this case denotes finding the mean value
of the prices and the departure from that mean. The trader will attempt to
enter the market when the price moves away from the average established,
expecting it will return to the same. This, to a certain extent, is the same as
following an uptrend, buying as the trend falls towards the low retracement
line and selling as it nears the high part of the channel (or selling short).
To end this brief digression, we could recap by saying that trend trading
provides for breakout entry points with stop orders, while countertrend
trading attempts to enter the market with limit orders placed on retrace-
ments. EUR/USD, GBP/USD, and EUR/JPY are usually valid trend
trading currency pairs, while USD/JPY, GBP/CAD, and EUR/NZD are
better suited to countertrend trading to name but a few, without wishing to
influence the reader’s trading, but, rather, simply summing up the author’s
experience in terms of the studies conducted so far.
Setting aside the advantages associated with diversification, let’s look at
the advantages strictly in terms of the subject of this book, money manage-
ment.
279
In the previous pages, we described the effects of applying a
APPENDIX III
position-sizing algorithm to various strategies, and it’s evident that
this can emphasize the effects of a winning strategy considerably.
I must once again underline, to avoid any misunderstandings, that money
management will not turn an approach that’s basically a loser into a win-
ner. To trade on the markets, you need winning strategies, and a suitable
money management algorithm can maximize the benefits. A losing strategy
will remain such, with the exception of some isolated cases.
Whatever the model applied, the basic principle is to scale the position
on the basis of the level of capital reached. As already mentioned, the num-
ber of trades helps make position sizing more effective, as one can act more
frequently and therefore follow the evolution of the capital ‘from closer at
hand.’
Let’s suppose we’re trading EuroFX futures; obviously changing exposure
means a move of one contract at a time, and for every new contract we add
(or subtract) we should wait for the relevant change in terms of capital.
Perhaps, the change from the first contract to the second and then the
third will take quite a long time as it may be necessary to accumulate a
considerable amount of revenue. Before adding the second contract, all the
trades will be made using the same ‘multiplier’ equal in fact to 1, as only 1
contract can be used. So, let’s suppose we’re using a winning strategy. The
effect of the trades between one step and the next will therefore be limited
in terms of the possibility to be amplified as this is linked to the step-by-step
increment to which market exposure is subject. This is the same practical
obstacle that emphasised the purely theoretical aspects of optimal f calcu-
lated in a continuous environment in which we were able to use infinitesimal
fractions of a contract, while in reality obviously this cannot be done.
Now considering the Forex market, we have to be aware that one EuroFX
futures contract is the equivalent, in dollar move terms, to an exposure of
125,000 USD in EURUSD. This amount doesn’t correspond to the standard
lot, which is €100,000, but it can in any case be used due to the fact that
most brokers offer the option of trading not only with standard lots but also
with mini lots (€10,000) or micro lots (€1,000).
At this point, it’s evident that, whereas before we had to wait for a certain
increment in order to add a new futures’ contract, in this case we can use a
variety of intermediate lots to follow the equity trend much more closely.
To better analyze this characteristic, a series of random trades on Excel has
been created, corresponding to a hypothetical series of results for a EuroFX
280 futures trading system. The equivalent result on Forex was then calculated,
APPENDIX III
in the hypothesis of trading with one standard lot (for example, a result of
250 USD in the first case would produce 200 USD in the second, while 150
USD in futures would be the equivalent of 120 USD in Forex).
This series was used as a reference to which a position sizing algorithm
was applied, in particular percent f model set at 0.5% on an initial capital
of 100,000 USD. If we make a hypothesis of various possibilities in terms
of scalability of the lots in Figure AIII.1 we can compare the final results
after 100 trades depending on whether we want to trade EuroFX futures
with standard lots of €100,000, mini lots of €10,000 or even micro lots of
€1,000.
It’s immediately clear that even using €100,000 lots produces advantages
over futures’ contracts (which would be the equivalent of a €125,000 Forex
exposure). The real game-changer, though, is the possibility of scaling the
positions in multiples of €10,000, and there’s another minor improvement
if micro lots are used.
The action described in this appendix is very similar to what we said about
the number of trades, which explained how money management models
375,000
EuroFX Futures
100k lot
10k lot 337,589
1k lot 335,233
325,000
310,507
302,044
275,000
225,000
175,000
125,000
75,000
FIGURE AIII.1 Numerical results of the application of the same position sizing model
to different trading scenarios in terms of lots.
become more effective the greater the number of changes made to the num-
ber of contracts. Also in this case, using smaller lots is the same as mak-
ing a greater number of changes. In fact, you don’t change from 1 to 2 in 281
APPENDIX III
1 go, keeping the first contract going for many trades; rather, there’s an
almost-continuous change, trade after trade, made possible by the infinites-
imal steps we can now have.
Considered in these terms, the Forex market is certainly interesting. An
extremely dynamic management of the position maximizes the results of
our efforts to apply an effective money management model. In this case, the
Forex market is the market par excellence.
The same reasoning goes naturally also for CFDs, which, depending on
the instrument they’re replicating, offer various scalability options. It’s advis-
able, when dealing with these instruments, to consider the impact of the
spread in the calculations, as often it’s significant (not by chance this is where
the brokers offering the instrument make a profit) and this may have a huge
effect on final results.
APPENDIX IV
Online Trading
■ IV.1 The Trader
We’ve discussed the risks associated with trading, how to try to avoid the
worst-case scenarios, and concentrated on various techniques that can be
used to effectively size a position. All of the above are made up of a cer-
tain dose of science, while attempting to juggle the figures to avoid being
overwhelmed.
282
But the risks don’t end there. Learning the notions in this book isn’t
enough to trade lightheartedly on the markets; there are other perils to
watch out for.
Figure AIV.1 shows an example of the risks in the trading world.
First and foremost, one might ask, what is trading really? Someone who’s
new to this world will often have the wrong idea after watching a film, read-
ing an article, or perhaps end up falling prey to misleading advertising.
In terms of films, I must say the characters portrayed, while it’s true they
may certainly exist, are nothing like this trader or many others I know.
The greedy, unscrupulous banker, always wearing braces and chewing on
a huge cigar, belongs to a different world than the reality we live in on a daily
basis. Anyone who starts trading on the stock exchange with the idea of one
day becoming that sort of character, perhaps looking for the same sort of
power to wield, would do better to choose another career. They’d probably
get as far as buying a pair of red braces to hold up their pants.
So, all intentions of becoming the next Gordon Gekko aside, the aspiring
trader does, however, have to take moral scruples into account; some people
might offer the reproach that traders don’t actually produce anything, oth-
ers might accuse them of exploiting the gullible. One might make the first
Financial
ignorance
Technological
Operational risk Broker risk
risk
Market risk
FIGURE AIV.1 Trading risks.

283
APPENDIX IV Online Trading

objection about many other professions too, but apart from this, it’s just not
true that traders are ‘of no use’ to society. A trader’s role on the markets, as
much as it may be just a drop in the ocean compared to the overall dimen-
sions of the same, is still one piece of the puzzle, and if various pieces are
lost, the puzzle will never be complete. No matter how small, an individual
trader’s contracts put on the market contribute to the market’s liquidity. This
brings us to the second objection. Speculation is seen as something bad, and
it would be if the trader effectively manipulated the prices for personal gain,
but what traders actually do is enter market movements, to become part of
the same, and as a part of these movements they also limit the range due to
the greater added weight, as a result making it more difficult for the market
to be manipulated by anyone who might actually try to do so. Obviously, one
trader’s contribution is but a grain of sand, but a beach is formed of many
grains of sand, and removing them one by one will eventually change that
beach forever.
Liquidity is fundamental for a market and lies at its very heart. On com-
modities’ markets it’s essential to offer to those that produce physical goods,
at a fair price, the opportunity to buy instruments to cover the risk. If the
markets didn’t offer a full range of said opportunities, it wouldn’t be the
trader who suffered, but the manufacturer first and foremost.
To put it simply, one might imagine the trader represents an obstacle to
the buyer who wants to speculate on the price of soya, or the seller who
wants to do so with gold futures.
These are perhaps somewhat extreme concepts, but they do represent the
real situation. We shouldn’t forget that those who criticize often do so out
of envy rather than to be constructive towards society.
■ IV.2 Trading Profits

This might be a sore point for some; unfortunately, trading for ‘the masses’
is no longer driven by the markets, it’s driven by the trading market. This
market has everything more or less anyone might need to start trading
online.
One starts with education and then brokerage, and it’s all absolutely legit-
imate. I opened the Unger Academy to pass on what I know about automatic
284 trading. The problem is with the messages related to the services offered.
In order to tempt the interested party (or arouse their interest), most
appeal to one of the most common traits found in humans: greed. Some
also play on fear but, in general, this is less harmful; it can often produce
results but it’s not as effective as dangling the prospect of riches in front of
someone’s nose.
If you look at the ads for online trading, they all play on the concept of
making money, and a lot of it! Most ads promise you’ll earn mind-boggling
figures with very little effort, very quickly, and above all with a ridiculously
low capital.
Is it really possible? Yes it is, in the same way you can buy a lottery ticket,
but if I told you to buy a lottery ticket because I once won the jackpot years
ago, not many people would do so, would they? Obviously not, because it’s
perfectly obvious how improbable that win is, and that luck is the only real
reason for such success. In trading, this doesn’t happen. The world of trading
has always appeared different, so people believe the dodges proposed by the
cat and the fox, and throw themselves into the fray. I use that term because
it’s what most people do, but believe me when I say, at least if you’ve bought
a lottery ticket you know how much you’ll likely lose – the price of the
ticket, no more.
The fairytale of the home trader who takes a break from his/her daily
routine to grease the wheels of the trading account is nothing other than a
fable. It’s something that can’t really exist, and it only does in the mind of
people who want to keep believing in such fairytales.
Risk takes the stage, disguised as an opportunity; but it’s a false opportu-
nity because all it does is eliminate some unfortunate person who dabbled in
trading blinded by a mirage called get-rich-quick. This person will never be
a trader: first and foremost because they’ll lose everything, but also because
at the end of the day, they’re not that interested anyway.
In the book there have also been cases of interesting performance, but
always at considerable risk, and it’s been said many times that these cases
shouldn’t be taken as examples of profit but only as far as the technicalities
of money management are concerned. This is because any example made
obviously refers to the past, and strategies that worked in the past may no
longer work in the future, or in any case may no longer work sufficiently
well to meet the requirements of a certain position sizing algorithm. We’ve
seen how money management can also amplify the negative phases, and it’s
easy to imagine where that might lead.
285
So is it impossible to make money on the stock exchange? I wouldn’t say

that. I would dampen overenthusiasm that comes from believing in foolish
advertising, but with serious hard work you can get a considerable amount
of satisfaction from trading, and this is the sore point, serious hard work. In
other words, you can’t all of a sudden become a trader. You need more than
the eager trader’s manual on how to make money, you need to study with
dedication and perseverance.
Then what? When will you get good at trading? I can’t provide certain
figures but, based on my experience over the years in direct contact with
other ‘real’ professionals in the sector, I’d say a good trader using a trust-
worthy trading system infrastructure could make something like 30% per
year, and I’m talking about home traders, not the Goldman Sachs broker
where the masses can represent a further obstacle.
Not much? Too much? My vote goes for the second, although many might
think it’s not much at all. For that matter, if we consider different initial capi-
talisation situations, it’s obvious you won’t get rich from one day to the next.
In reality, the profits mentioned can’t be considered a negligible per-
formance and remain valid for a good trader. What’s more, this kind of
performance can be obtained with a moderate risk: in other words, not a
percent f model set at 5%, but much less. Of course, everyone is free to
choose their own exposure and raise that number as high as they want; but
the risks increase greatly and often this leads to trouble and nothing else.
I also mentioned trading systems, yes, I’m a systematic trader. Here at the
Unger Academy I teach people how to build trading systems, but in the book
there’s also an example of discretionary trading. So, do I know about that,
too? Personally, I wouldn’t say I’m great at discretionary trading. I would say,
though, that if we compared the best systematic trader in the world with the
best discretionary trader in the world, the second would come out on top.
So why am I a systematic trader? Simply because I don’t think I’ll ever be the
best discretionary trader in the world, or even come close, but I do know
how to build trading systems.
■ IV.3 Systematic or Discretionary?

At this point, now that I’ve said I prefer systematic trading, it’s only right I
offer an explanation.
I prefer systematic trading first and foremost because it’s repeatable and
286
measurable; there are no hypotheses that aren’t corroborated by figures.
Once a trading system has been built, that’s it, and there’s no room to accuse
the user of psychological failings or anything else when accumulated losses
have him/her cornered. Discretionary trading is in reality too simple a field
for real course crooks to be a big success. These people are often very good
at using psychological tricks to play the student on the hook, and convince
him/her the losses are caused by the approach adopted, and not by the
self-proclaimed guru’s teachings. It goes without saying that these ‘gurus’
never take the trouble to prove the worth of their techniques on the actual
market.
I’m not saying this for the purpose of controversy, but simply to warn
novices against this type of risk, which is treacherous to say the least.
I certainly don’t want to insinuate that all those who teach discretionary
trading are crooks; on the contrary, I know a few who are really good traders.
What I can say though with conviction, is that while systematic trading,
consisting of rules, can be taught, this isn’t always true for discretionary
techniques, which involve a human component that can’t always be passed
on to others.
Remaining within the scope of money management, you’ll have seen how
many simulations were run and how many calculations done to evaluate the
various techniques. This was possible because we had the figures for the
approaches in question. We would never have been able to make particu-
larly detailed plans based on discretionary rules without a track record to
give us at least an idea of how advantageous said technique was.
Systematic trading also gives the trader greater freedom. The fact that
most systematic traders remain glued to the monitor studying charts of
external data flows and develop systems put on the market in real time is
another question; one should be able to dedicate one’s time to something
else once the systems are validated for use, simply guaranteeing valid
supervision to make sure everything is functioning as it should. With
discretionary trading, this is often impossible.
Creating a technologically valid infrastructure involves those technolog-
ical risks shown in Figure AIV.1. If you don’t have the right software or a
computer that’s up to the job, you run the risk of an unexpected interrup-
tion in trading which, in general, means unwelcome losses (imagine a missed
stop-loss or reverse orders not executed on the market in highly volatile
conditions).
287
■ IV.4 Choosing the Broker

When talking of risks, one must of course include the broker. It’s not really
a question of the risk the broker could go bankrupt and our accounts end up
empty, although this can happen and sometimes can’t really be avoided (it
happened to me with PFGBest). It’s more a question of staying away from
real crooks who, enticing clients with extremely interesting conditions, end
up draining the financial resources of the victim like a disease drains your
energy.
Often, there’ll be a consultant who tells you which trades to make,
encouraging you to go too far in terms of exposure. Sometimes things
will work out (purely by chance), but more often than not, things go
from bad to worse. At this point, the consultant suggests (or openly asks)
you to pay more money into your account to either make up for losses
or unblock positions in difficulty. This continues until the client gives up
(usually after heavy losses) or, if the client wants to close the account after
making a profit, the consultant miraculously disappears, and this person
who previously wouldn’t leave you alone suddenly won’t even answer your
messages!
These swindles are set up by various brokers based abroad, and it’s all
but impossible to bring them to justice, so these organisations con people
without really risking a great deal; once again, though, everything is based
on exploiting greed or the belief you can make cash quick.
When you choose a broker, you will have usually already chosen which
market you want to trade on. Many brokers only offer their services on OTC
(over the counter) markets like Forex and CFDs in general. These aren’t
regulated markets though, and this can mean additional risks. I’m not say-
ing you need to be particularly worried about them, but there may be a
catch. In January 2015, the sudden movement, after the decision taken by
the Swiss National Bank to remove its 1.20 CAP on the euro, caused quite a
few problems also in relation to the possibility of brokers ‘adjusting’ contract
prices. It wasn’t exactly an illegal decision but, in some cases, was clearly not
advantageous for traders.
Apart from this case, it’s important to remember that many brokers also
act as market makers with these instrument, so traders fight both against
market movements and against the brokers. Once again, while I certainly
don’t want to insinuate that brokers are dishonest, it’s evident that one can
288 in some ways be considered to be at a disadvantage.
■ IV.5 Which Platform?

Which platform to use to build your trading systems (if this is the route
you’re taking) is important, both in terms of technological evaluation and
the choice of the broker.
Without a doubt, the most widely used software is MetaTrader, which is
extremely popular thanks to the fact that it’s offered for free by almost all
brokers you can trade Forex and CFDs through.
Opening an account with one of these brokers will give you immediate
access to the platform where you can program your own systems. The broker
also provides you with the real-time data flow to trade with. While all this
might seem wonderful, it’s important to stress there are some obstacles.
The first is the programming language used in MT4 (the most widely used
MetaTrader version), which can be confusing at first and isn’t exactly easy
to use. The other obstacle is that the available historical data are often too
limited to check the validity of the system developed on the basis of past data.
While you can solve the first problem by taking a specific MQL course
(which, however, still isn’t that easy to pick up quickly), the historical data
problem is slightly more complex. Online you can also find free databases,
but they’re not that easy to import and, if you do manage to do so, this data
isn’t the same as that of the broker. As these are OTC markets, the prices
may change from one broker to the next – not by much, obviously, but what
might work with one broker may not produce results with another.
If you decide to trade in futures (or on the stock exchange), you’ll need
another type of software such as MultiCharts or TradeStation, NinjaTrader,
ProRealTime, AmiBroker, and others.
MultiCharts uses practically the same programming language as TradeSta-
tion (called PowerLanguage on MultiCharts and EasyLanguage on TradeSta-
tion), and it’s quite easy to learn, but the data must be acquired from external
suppliers, which means additional costs. The platform itself isn’t the cheap-
est, but does offer a 30-day free trial. TradeStation is also a broker and,
depending on the offer, may be advantageous to minimize costs in the case
of a certain amount of trading costs per month with the platform. In any
case, opening an account with a minimum capital of a few thousand euros
may be the best solution for those who want to start investing just a small
amount.
289
The same can be said for NinjaTrader, which is programmed using C#, a

real programming language, and this can represent a significant obstacle.
ProRealTime offers various advantages such as simple language (similar to
that used in MultiCharts) but also has various limits both in terms of the com-
plexity of the systems that can be programmed and in terms of the available
data limited to what’s supplied (intraday for a fee) that can’t be edited – for
example, importing ASCII data (not supported).
AmiBroker is very versatile but limited to real time, and the initial learn-
ing curve can be steep.
Finally, I’d like to emphasise that you can become a trader, but it should
never be considered a way out of a difficult economic situation. Trading isn’t
a solution for just about anyone who’s unemployed, or a cure-all for all the
economic problems a big family might run into. It’s a good idea to bear this
in mind to avoid turning your dream into a nightmare. Trading is a very
demanding profession, which can be learned with study and self-sacrifice
just like any other. No one would dream of opening a dental studio without
knowing the first thing about it, just because dentists earn a lot, but many
new traders approach trading with this idea.
INDEX
2% risk, 87f 20-day ATR, 201–202

3% risk per trade, adoption, calculation, 205, 222, 242t
187 consideration, 82
4% risk, 149f–152f EasyLanguage code, 191,
5% fixed risk percentage, 202–203, 209
application, 74f method, 200–201
10% risk, impact, 75f percent risk with ATR method, 291
200–201
A pre-entry level, 241f
Aggressive delta, impact, 107f trading, 240
Aggressive ratio, 97–98 usage, 86
Allowed volatility, usage, 88f
Anti-Martingale, 1, 9–14 B
approach Back testing, 115
power, 15 Ball, picks
usage, 11 anti-Martingale system, usage,
ball, picks (examples), 15f–16f 15f–16f
coin tosses, examples, 10f, Kelly formula, usage, 28f, 30f,
12f–14f, 18f–20f 31f
Asymmetric ratio, 99–100 probability, 16
Average drawdown, calculation, winning, probability, 27
144 Base percentage, reduction, 103
Average true range (ATR), 225, Bet, increase, 4f–7f
226t Boundary, 144
Brokers analysis, 17–20
leverage, 187 anti-Martingale system,
selection, 287–288 examples, 10f, 12f–14f,
18f–20f
C bet, increase, 4f–7f
Capital consecutive losses, 7t–8t
amplification, 39f, 44 distribution, 11
availability, 76 Kelly formula, usage, 23f,
exposure, 186f, 188f, 189f 25f–27f
reduction, 49f losses, 6
funds, withdrawal, 248 luck, 12
increase, 94–96, 249 probability, calculation, 8
increment, production, 29 results, 24
levels, 70f series, test, 17
liquidity, addition, 248 success rate, 11
loss, 61 Commissions
management, 194 impact, minimization, 187
production, 96 payment, 186
reduction Congestion channel, 218
accumulated losses, Constant monthly income, trading
292
impact, 48 (impact), 251
INDEX
risk, impact, 87f Contracts

risking, 64 basis, examples, 92t
risk, result, 117t calculation, 40, 50
selection, 169–173 ff contracts, calculation, 235, 242
stake, 3–4, 6 gain per contract, contract
usage number increase (ratio), 69
example, 2 Kelly formula, contrast, 41f
stake, establishment, 1–2 loss, 67
volatility, limitation (results), number, calculation, 172,
84f, 85f 220–222, 226–227
Capital of reference, 225–228, formulas, usage, 67, 72, 248
243t percent f calculation,
Cash flows, size (increase), 94–95 EasyLanguage code, 262f
CFDs, 281, 288 single contract strategy, Kelly
Classic % f line, fixed ratio line formula (contrast), 40
(comparison), 99f trading trend line, comparison,
Coin tosses, 1–2 101, 102f
Contracts, numbers usage, 79f, 98t, 166f–167f
calculation, 101 example, 77–78, 99
change, levels, 91–92 Discretionary trade (example),
usage, 70f DAX futures (impact), 228,
Core equity model, 170–171, 196 229f
reduced core equity model, Discretionary trading, 215
170–172 Mediaset stocks
Countertrend strategy example, 218–225
code, 269f figures, 219f
performance report, 269f preference, 286–287
Crabel, Toby, 253 Diversification, increase, 198
Crude Oil System Diversified systems, 60
annual period analysis, 159f Donchian channel breakout,
equity line, 159f 157–158
fixed ratio method, usage, 165f Drawdown (DD), 98. See also
percent volatility, 165f Percentage drawdowns
strategy performance report, 95% drawdown, 247
158f calculation, 144
system performance report, 165f data, 147f
‘‘Cut your losses and run’’ rule, 216 distribution, 128, 133
293
data, 148f, 150f
INDEX
D fixed ratio method, usage, 134f
DAX example, 47f
contract, 182 expected drawdowns
points, 235 risk, impact, 151f
futures, impact, 228, 229f values, risk (impact), 152f
trade, 244 final value, trading (impact),
Delta, 174, 244 155f
acceleration, 70 impact, 194
aggressive delta, impact, 107f importance, 249
change, 89f levels, 58f
consideration, 78 maximum historical drawdown,
fixed ratio, 106f–107f 70
method, usage, 127f–130f, maximum percentage/absolute
133f–134f, 147f, 175f drawdown, 59
impact, 165f, 177f maximum percentage
maximum theoretical loss, drawdowns, 118f
relationship, 79f–81f percentage (DD%), basis, 144
Drawdown (DD) (Continued) reduction, 76
percentage of vol (%vol), impact, stop, presence/absence,
109f 105f–106f
real maximum loss, usage, 138f strategy development, 261f
reduction, 107f, 153 timid bold equity, 100–102
risk total equity model, 170, 196
basis, 57–58 total equity per position
confirmation, histogram allocation, 204f
(usage), 128 trade, relationship, 42t
impact, 138f, 146f Euros stop-loss per contract, level
significance, 77 (performance report/equity
values, 122f line), 264f, 266f
Eurostox50 futures, 268
E Euro Stoxx 50 futures, trading
EasyLanguage code, 83, 209, 253, (example), 253
289 Exit level
contracts calculation, 262f increase, 233f, 234f
days, exclusion, 257f specification, 246
example, 35f, 183f Exit point, increase, 231f, 232f
exposure calculation, 261f Expected drawdowns
294
percent f calculation, 262f values, risk (impact), 152f
INDEX
strategy, 254f Expected drawdowns, risk

writing, 248 (impact), 151f
Equally weighting positions, 196 Expected returns, risk (impact),
Equity 152f
core equity model, 170, 196 Exposures, exposure per position,
curve trading, 103–110 210f, 211f
decline, 250
drawdown levels, 58f F
final value, trading (impact), f (fraction of the capital risked), 54,
155f 61. See also Optimal f;
increase, winning trade (impact), Percentage of f; Reduced f
59 loss, 63
peak equity values, Kelly formula ff. See Fixed fractional
(application), 46f Filters
reduced total equity model, ADX filter, 210f–214f
170–172, 196 usage, 250
Fixed fractional ( f f ) contracts, FR. See Fixed ratio
calculation, 73, 136, 235, Funds
242 management, 193–194
Fixed fractional ( f f ) method, withdrawal, 248
54–60 Futures
comparison, 101, 102f portfolio, usage, 251
hypothetical max, setting, 75 volatility, 240
risk, 85 Futures, trading, 228–244
limitation, 55, 56t entry level, movement, 239f
percentages, 55, 57f monitoring, 242t, 243t
trades, example, 55t scenarios, 236
usage, 66, 78, 94, 135f
Fixed ratio (FR) G
contracts, calculation, 73 Gain, calculation, 222, 235, 238,
delta value, example, 79f 242, 244
drawdown, impact, 107f Gaussian distribution, 111
equity, improvement, 106f–107f Global exposure, limitation, 251
line, classic % f line Gold System
(comparison), 99f annual period analysis, 161f
Fixed ratio (FR) method, 68–81 equity line, 160f
295
95% drawdown, 247 futures
INDEX
application, 174–175 fixed ratio method, usage, 166f
comparison, 100f percent volatility, 167f
delta, impact, 165f, 177f system, performance report,
drawdown trend, 128f 166f, 167f
increment level, 71 futures market, position sizing
performance report, 166f, 167f models, 166
preference, 249 future, strategy performance
usage, 114 report, 160f
examples, 127f–130f, 133f,
134f, 147f H
portfolio performance report, Histogram
175f confidence, 130
Fixed risk percentage, application, impact, 131
74f usage, 121–122, 124
Foreign exchange (Forex) trading, Historical results, obtaining,
advantages, 278–281 109–110
Holding period return (HPR), M
61–63, 62f Market volatility
values, 64f ATR measure, 86
multiplication, 62 decrease, 84
Martingale, 1
I approach, 5, 9–10
Intraday bar chart, comparison, 10–11
usage, 253 bet, increase, 4f–7f
Intraday data, usage, 183 system, 2–10
Intraday strategy, usage, 268 Maximum allowed drawdown,
155
J Maximum allowed percent
Jones, Ryan, 68–69 95, volatility, change, 90f
99, 247 Maximum historical drawdown,
70
K Maximum historical loss,
Kelly formula, 21 67, 120
application, 37–52, 46f Maximum loss, 35, 135–139
ball, picks, 28f, 30f, 31f possibility, 50
basis, 29–31 real maximum loss
296
coin tosses, examples, 23f, drawdown trend,
INDEX
25f–27f impact, 137f

drawdown, example, 47f risk, impact, 137f
equation/calculation, 22 Maximum percentage drawdowns,
results, problems, 40, 42 118, 174
single contract strategy, contrast, Maximum theoretical loss, delta
40 (relationship), 79f–81f
theoretical results, comparison, MaxLosing input, 261
43f Mediaset, discretionary trading
usage, 67, 77, example, 218–225
248, 249 Mini futures, usage, 90
Kelly percentage, usage, 42 Minimum risk percentage, usage,
77f
L Monetary stop-loss, 268
Limit risks, approach, 39–40 Monetary stop value, 261
Liquidity Money management
absence, 70–71 absence, 163f–164f
addition, 248 method, capital levels/contract
importance, 283–284 numbers, 70f
models, 53 Optimal f, 60–65
application, 164 calculation, 61
portfolio money management, results, example, 65f
168–169 TWR trend, 64f
effects, 173–180 TWR values, 64f
principles, application, 62 usage, 67, 77f
questions/answers, 246 Optimal percentage, 27
single systems approach, application, 101
avoidance, 169 Optimal value, representation,
strategies, 246 254–255
application, 143
types, understanding, 268 P
techniques, refinement, 94 Payout, Kelly formula (usage),
usage, 181, 247 22–23
Monte Carlo simulation, 114, 135 Percentage drawdowns
capital risk, result, 117t distribution
maximum loss, impact, 136 fixed ratio method,
numbers, permutation, 118t usage, 129f
trade distribution, change, 116f risk, impact, 132f
trade order, change, 116f fixed ratio method, usage, 130f
trade sequence, 115t 297
fixed ratio method
INDEX
usage, 114 (usage), 148f
Moving average values, comparison,
acceleration/deceleration, 34 273–274
basis, 33–35 Percentage of f (% f ), 54
calculation, 103 equity stop, 105f–106
crossover, 104f, 132, 208 equity trend, 104f
strategy, usage, 145 Percentage of vol (%vol), 108f
delta/stops, contrast, 107f impact, 109f
Multiplication factors, 5 Percentage profits, log-normal
distribution, 145
N Percentage returns
‘‘New Interpretation of Information comparison, 88f
Rate, A,’’ 21 distribution, 121f
fixed ratio method,
O usage, 133f
Online trading, 282 risk, impact, 131f
Opening Range Breakout principle fixed ratio method,
(Crabel), 253 usage, 127f, 147f
Percent f, 165f sizing model, application
equivalence, 263f (results), 281f
impact, 166f, 167f, 173f PowerLanguage on MultiCharts,
results, 263f 289
Percent f model Probability, calculation, 8
performance report, 265f, 266f Profits
portfolio equity line, 273f maximization, 21, 38, 53, 60, 90
value, setting, 264f, 266f, optimal value, 90
270f–272f optimal percentage, application,
Percent of equity method, 197 101
Percent risk, 198 percentages, reduction, 138
usage, 223 planning, 251
Percent risk model, 67 results (three-year database), 146f
Percent risk with ATR method, trading profits, 284–286
200–201 trend, 153f
Percent volatility (%Vol), 81–91, risk, impact, 149f
205, 217 stop-loss change, 259f
application, 238 trading time window basis,
impact, 206f–208f 256f
percentages, usage (example),
298 R
86f
INDEX
usage, 87f, 201 Range, calculation, 221

Percent volatility (%Vol) model, Real maximum loss
81–91, 205, 217 drawdown trend, impact, 137f
application, 238 risk, relationship, 137f
impact, 206f–208f usage, 138f, 139f
percentages, usage (example), Reduced core equity model,
86f 170–172
usage, 87f, 201 Reduced f, 95–97
Platform, selection, 288–289 delta, usage, 98t
Portfolio Management Formula method, steps (selection/results),
(Vince), 60 95t, 96f, 97t
Portfolio money management, Reduced total equity model, 223
168–169 Risk
effects, 173–180 basis, 57–58
Position calculation, 235
risk, 247 drawdown values, 122f
fixed fractional method, usage, drawdown values, 122f
135f examination, 3
fixed risk percentage, application, fixed ratio method, usage,
74f 129f
impact, 87f, 131f–132f, 146f, percentage drawdown
149f–152f distribution, fixed ratio method
increase, impact, 270 (usage), 129f
limitation, 55, 56t trend, 124f, 127f
principle, 66 percentage returns, fixed ratio
percent volatility (%Vol), usage, method (usage), 127f
87f results, production, 16
profile, 69 returns distributions, 123f
selection, 91, 145 returns trend, 126f
profits, results, 146f success return, 11
real maximum loss Single contract strategy, Kelly
relationship, 137f formula (contrast), 40
usage, 138f, 139f Soybeans System
reduction, 132, 239f annual period analysis, 162f
risk per trade, 186f, 188f, 189f equity line, 162f
trading risks, 283f futures 299
Riskiness, 68 fixed ratio method, usage,
INDEX
Risk percentage, 65, 75f, 76f, 271 167f
change, 89f system, performance report,
impact, 134f 167f, 168f
increase, 236 strategy performance report,
minimum risk percentage, usage, 161f
78f Standard deviation
returns, distribution, 121f increase, 144
Robbins, The, 247 tolerance, 154
Roulette, playing (simulation), 8–9 Stock market, trading, 181–191,
195
S Stocks
Secure f, 65–68 Mediaset, discretionary trading
Simulations example, 218–225
creation, 142 portfolio, usage, 251
drawdown trend, fixed ratio percent volatility model,
method (usage), 128f impact, 207f
Stocks (Continued) money management, absence,
trading, money management 163f
(usage), 181 percent f model, value (setting),
volatility, measure, 82, 199 271f, 272f
Stop-loss result, 272f
2% stop-loss, impact, 183
change, 259f T
hypothesis, 237f Terminal wealth relative (TWR),
level, 230, 236 61–63, 62f
setting, 57 trend, 64f
tightness, 217 values, 64f
triggering, 199, 216, 222 Time stop, usage, 260
absence, 38f, 50, 52 Timid bold equity, 100–102
occurrence, 55 mixed curve, blocking, 103f
usage, 136, 139 Total capital, entry level
Stop, movement, 224f (movement), 239f
Stop percentage, calculation, Total equity model, 170, 196
218 reduced total equity
Sufferance, 85 model, 196
Systematic trading, preference, Traders
300
286–287 risks, 282–284
INDEX
Systems. See Anti-Martingale; temperament, importance,

Martingale; Trading system 94–95
bankruptcy, 86 Trades
blocking, 109 allowed volatility, usage, 88f
combination, usage, 157–168 avoidance, 246
improvements, 208–214 contract basis, examples, 92t
results, 214t days, exclusion, 258f
irregularity, 105–106 dependence. See z-score.
questions/answers, 246 dependency, belief, 250
results, percent f model value distribution, change, 116f
(setting), 272t entry, 183f
Systems portfolio equity, relationship, 42t
annual trend, money example, 55t
management (absence), 164f expected trades, risk (impact),
equity line, 270f 151f
money management, absence, history, creation, 119
163f HPR values, 64f
losses, 61 criteria/definition, 215–218
money management study, 246 discretionary trading, 215
order, change, 116t equity curve trading, 103–110
parametrization, 60 foreign exchange (Forex) trading,
percentage drawdown trend, advantages, 278–281
fixed ratio method (usage), futures, trading, 228–244
148f impact, 154f–155f. See also
performance summary, 36f Constant monthly income.
profits trend, risk (impact), online trading, 282
149f platform, selection, 288–289
risk percentage, 75f preference, 286–287
sequence, Monte Carlo profits, 284–286
simulation, 115t risks, 283f
stake, establishment, 1–2 scenarios, position sizing model
stop-loss, triggering (absence), (application results), 281f
38f, 50, 52 steps, selection/results, 95t, 96f,
theoretical trades, consideration, 97t
250 techniques, refinement, 94
usage, cessation, 250 Trading Game, The (Jones), 68
volatility, adjustment, 225–228 Trading system, 33, 252–267, 286
analysis, moving average basis, 301
TradeStation strategy
INDEX
equity curve, 39f 33–35
performance, contract/Kelly combination, usage, 157–168
formula (contrast), 40, 41f dynamics, example, 34f
trades, example, 38f EasyLanguage code, 35, 254f
TradeStation strategy performance impact, 252
report improvements, 208–214
annual strategy results, 37f Kelly formula, application,
strategy report, 36f 37–52
Trade Your Way to Financial Freedom strategy report, 36f
(Tharp), 82 True range. See Average true range
Trading usage, 199–200
5% risk strategy, 247
aggressive trading V
strategy, 247 Van Tharp, K., 82, 101, 170
cessation, 70–71, 106, 109 Vince, Ralph, 60–61, 66
continuation, minimum risk Volatility
percentage (usage), 78f adjustment, 225–228
Volatility (Continued) Winning, probability, 22–24, 27,
allowed volatility, usage, 88f 29, 37
limitation, results, 84f, 85f Work plan, 141
market volatility, decrease, 84 method, contrasts, 251
maximum allowed percent usage, 141–155
volatility, change, 90f Worst-case scenario (WCS),
percent, impact, 165f 61–63
period volatility, 82
Z
W z-score (trade dependence),
Williams, Larry, 60, 66–67, 249 110–112, 144
Win/loss alternation, 111 usage, 250
302
INDEX

Unger A. - The Successful Traders Guide To Money Management (Wiley 2021)

Uploaded by

Copyright:

Available Formats

Unger A. - The Successful Traders Guide To Money Management (Wiley 2021)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Unger A. - The Successful Traders Guide To Money Management (Wiley 2021)

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What strategies for money management does it discuss?

What strategies for money management does it discuss?

THE SUCCESSFUL TRADER’S

© 2021 John Wiley & Sons, Ltd

Library of Congress Cataloging-in-Publication Data is Available

ISBN 978-1-119-79880-4 (hardback)

Cover Design: Wiley

Set in 12/14pt, PerpetuaStd by SPi Global, Chennai, India

CHAPTER 1 Martingale and Anti-Martingale 1

CHAPTER 2 The Kelly Formula 21

CHAPTER 3 A Banal Trading System 33

CHAPTER 5 Refining the Techniques 94

5.7 z-Score 110

CHAPTER 6 The Monte Carlo Simulation 114

CHAPTER 7 The Work Plan 141

CHAPTER 8 Combining Forces 157

CHAPTER 9 Money Management When Trading

CHAPTER 10 Portfolio Management 193

CHAPTER 11 Discretionary Trading 215

CHAPTER 12 Questions and Answers 246

APPENDIX III 278

T he second millennium began with the explosion of online trading in

for the next fad, all riding the wave of greed.

MARTINGALE AND ANTI-MARTINGALE

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

Martingale (increase bet after loss)

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

MARTINGALE AND ANTI-MARTINGALE

Martingale (increase bet after loss)

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

Martingale (increase bet after loss)

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

MARTINGALE AND ANTI-MARTINGALE

Starting with a higher percentage speeds up this process considerably, as

MARTINGALE AND ANTI-MARTINGALE

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

FIGURE 1.5 Loss 1, win 1.25 – anti-Martingale system.

MARTINGALE AND ANTI-MARTINGALE

(*) N.B.: In reality the capital isn’t mathematically lost as it is using

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

MARTINGALE AND ANTI-MARTINGALE

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

Martingale (increase bet after loss)

after 100 tosses after 1000 tosses

% risk ending capital gain % % risk ending capital gain %

MARTINGALE AND ANTI-MARTINGALE