Portfolio Theory and Behavioral Finance

(w.IN.23HS.V)
Chapter 2: Investors Choice and Utility
Asset Allocation and Diversification

Building Competence. Crossing Borders.

Dr.-Ing. Martin Schnauss, CFA, FRM / Dr. Jan-Alexander Posth

[email protected], [email protected], 2020 / Vers. 1.1.3
Learning Objectives Chapter 2

After this chapter you know:

• How to apply concepts as:
– Investors choice and utility
– certainty equivalent

– Risk and return objectives of investors

– The asset allocation process

– Leveraging of portfolios

Literature: Kritzman, Mark (1992): „What Practitioners Need to Know

about Utility“. Financial Analysts Journal. 48 (3), pp. 17-20.

-2-
Topic Investors Choice and Utility

Investors Choice and Utility

-3-
Utility Introduction

• Goal of the lecture: Introduction to utility theory and its

application to financial analysis.
• An important assumption in financial theory is that
rational investors seek to maximize expected utility.
• What about maximizing expected wealth? Consider the
famous
St. Petersburg paradox (first studied by the Swiss
mathematician Daniel Bernoulli in the 18th century).
– Suppose the following game is proposed: A coin is flipped
until head appears. You win $2𝑖𝑖 if the first head appears on
the 𝑖𝑖 th toss of the coin.
– How much would you be willing to pay for the game?
– What is the expected value of the game?
-4-
Utility Introduction (cont’d)

– Expected value of the game:

∞ ∞ ∞
1 𝑖𝑖−1 1 𝑖𝑖 −1 1
E 𝑋𝑋 = � 𝑖𝑖 2 = � 𝑖𝑖 2 2 = � 1
2 2 2
𝑖𝑖=1 𝑖𝑖=1 𝑖𝑖=1

2 3
0
1 1
1 2
1
=2 ⋅ +2 ⋅ +2 ⋅ +⋯
2 2 2 Probability to win 1’000
is approx. 0.05%...
1 1 1
= 1 ⋅ + 2 ⋅ + 4 ⋅ + ⋯ = ∞∞
2 4 8

– Nobody would pay an infinite amount of money for this

game because significant payoffs are not very likely.
– Bernoulli’s solution to the paradox was to argue that
people do not maximize expected wealth but expected
utility and that utility was increasing with wealth at a
decreasing rate. -5-
Utility as a Function of Wealth

• Example of a utility function 𝑢𝑢 𝑥𝑥 which is increasing in wealth and

exhibits diminishing marginal utility, that is
– 𝑢𝑢′ 𝑥𝑥 > 0 (marginal utility is positive, i.e. more wealth is
This is the slope of u preferred over less)
– 𝑢𝑢′′ 𝑥𝑥 < 0 (diminishing marginal utility, as wealth
This is the curvature (or increases, the increments to utility become
concavity) of u or the smaller)
slope of the slope of u

– Kritzman, Figure A
Utility is increasing with wealth
at a decreasing rate.
Slope decreases
from “blue” to “red”
 u’’ is negative

-6-
Utility as a Function of Wealth (cont’d)

– Kritzman, Figure B
A negative second derivative implies that we would experience
greater disutility from a decline in wealth than the utility we
would derive from an equal increase in wealth.

– People usually refuse to play fair

games. They prefer certainty.

-7-
Log-Utility

• Suppose utility is given by 𝑢𝑢 𝑥𝑥 = ln(𝑥𝑥)

– Verify that marginal utility is positive for every wealth level 𝑥𝑥 > 0:
𝑢𝑢′ 𝑥𝑥 > 0 ′ 1
ln 𝑥𝑥 = > 0∀𝑥𝑥 > 0
𝑥𝑥
– Verify that the function exhibits diminishing marginal utility:
′
𝑢𝑢′′ 𝑥𝑥 < 0 1 1
= − 2 < 0∀𝑥𝑥 ≠ 0
𝑥𝑥 𝑥𝑥
– Numerical example (Kritzman, p. 17)
− How much increases utility if wealth increases from 100 to
150?
Ln(100) = 4.60; Ln(150) = 5.01 ln 150 − ln 100 = 0.41
Δx=50 in both cases!
− If wealth increases from 150 to 200?
Ln(150) = 5.01; Ln(200) = 5.30 ln 200 − ln 150 = 0.29

-8-
Expected Utility

• An important assumption in financial theory is that

rational investors seek to maximize expected utility.

• What is the expected utility of the game considered before

if we assume a log utility function? (Also assume for
convenience that initial wealth equals 0.)
∞ ∞ ∞
1 𝑖𝑖−1
𝑖𝑖 − 1
𝐸𝐸 𝑢𝑢 𝑋𝑋 = � 𝑝𝑝𝑖𝑖 𝑢𝑢 𝑥𝑥𝑖𝑖 = � 𝑖𝑖 ln 2 = 𝑙𝑙𝑙𝑙(2) � 𝑖𝑖
2 2
𝑖𝑖=1 𝑖𝑖=1 𝑖𝑖=1
= ln(2) ≈ 0.69

-9-
Risk Aversion

𝑢𝑢′′ 𝑥𝑥 < 0

• If the utility function is concave (that is, exhibits

diminishing marginal utility), fair bets are avoided.

• Given log-utility, what is the expected utility in the

following cases
– $100 with certainty (initial wealth)?
– $90 with probability 0.5 and $110 with probability
0.5?
– $50 with probability 0.5 and $150 with probability
0.5?

- 10 -
Certainty Equivalent

• Definition: The amount of payoff (e.g. money or utility) that an

agent would have to receive to be indifferent between that
payoff and a given gamble is called that gamble's 'certainty
equivalent'. For a risk averse agent (as most are assumed to
be) the certainty equivalent is less than the expected value of
the gamble because the agent prefers to reduce uncertainty.
(Source: http://economics.about.com/)
• Given log-utility, what is the certainty equivalent of the
following gambles:
– $90 with probability 0.5 and $110 with probability 0.5?
ln 𝐶𝐶 = 0.5 ln 90 + 0.5 ln 110 Inverse of ln() is exp()
𝐶𝐶 = 𝑒𝑒 0.5 ln 90 +0.5 ln 110 = 99.5
– $50 with probability 0.5 and $150 with probability 0.5?
ln 𝐶𝐶 = 0.5 ln 50 + 0.5 ln 150
𝐶𝐶 = 𝑒𝑒 0.5 ln 50 +0.5 ln 150 = 86.6
- 11 -
Graphical Representation

𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈
ln(150) ln(𝑥𝑥)
ln(C)
ln(50)

86.6
50 100 150 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 (𝑥𝑥)
- 12 -
Risk Premium

• The certainty equivalent of $86.60 means that we would

be indifferent between getting $86.60 with certainty or
getting $50 with a probability of 0.5 and $150 with
probability 0.5.

• The difference in expected value that would just induce

us to choose the risky prospect over the certain outcome
is called the required risk premium. In our example the
risk premium is
0.5 ⋅ 50 + 0.5 ⋅ 150 − 86.60 = 100 − 86.60 = $13.4
or 13.4/86.6 = 15.47%

- 13 -
Risk Premium (cont’d)

• Assume , instead, that someone is starting with initial wealth of

$1,000, and ends up with $950 or $1,050, each with probability
of 0.5.
The certainty equivalent would then be $998.75 and the risk
premium $1.25 or 0.125%.
• The wealthier person is less risk averse in absolute terms
because the $50 gamble is just a small fraction of his wealth.
 Log utility implies decreasing absolute risk aversion.
• If the potential gain or loss would be $500 (also 50% of initial
wealth), than the certainty equivalent would be $866 and the
– risk premium = 1000- 866 = $134, or
– also 134/866 = 15.47% .
 Log utility implies constant relative risk aversion.
- 14 -
Willingness to Pay for Insurance

• Imagine an individual that owns $100,000 and there is a probability of

1% that he loses $60,000 (say by a fire in his flat) during the next
year.
• Further assume that his utility is logarithmic.
• Without insurance, the expected utility of this person will be
𝐸𝐸 𝑢𝑢(𝑋𝑋) = 0.99 ⋅ ln 100000 + 0.01 ln 40000 = 11.50376
• The fair insurance premium in our example would be
0.01 ⋅ 60000 = $600
• If the person gets complete insurance, his utility would be And that’s why
everybody is
𝑢𝑢 100000 − 600 = ln 99400 = 11.50691 happy to overpay
which is more than without insurance. for insurance…

• Up to how much money would this person pay for insurance?

ln 100000 − 𝒙𝒙 = 11.50376
Therefore he would pay up to $912.
- 15 -
Measures of Risk Aversion

• Economists distinguish between risk aversion, risk

neutrality and risk seeking. Empirical work shows that
most people are risk averse.
• Kritzman (p. 19) states that economists distinguish
between the absolute amount of one’s wealth that is
exposed to risk versus the proportion of one’s wealth that
is exposed to risk, and whether this amount decreases,
remains constant or increases as wealth increases.

- 16 -
Measures of Risk Aversion (cont’d)

• The more concave the utility function the higher the degree of risk
aversion.
• Important quantitative measures of risk aversion were introduced by
Arrow and Pratt.
• Measure of absolute risk aversion:
Remember slide 8!
𝑢𝑢′′ 𝑥𝑥
𝐴𝐴𝐴𝐴𝐴𝐴 𝑥𝑥 = − ′
𝑢𝑢 𝑥𝑥

Example: If 𝑢𝑢 𝑥𝑥 = ln 𝑥𝑥 , then 𝐴𝐴𝐴𝐴𝐴𝐴 𝑥𝑥 = 1/𝑥𝑥. Therefore, log utility

implies decreasing absolute risk aversion (as wealth increases).
• Measure of relative risk aversion:
𝑢𝑢′′ 𝑥𝑥
𝑅𝑅𝑅𝑅𝑅𝑅 𝑥𝑥 = 𝑥𝑥 ⋅ 𝐴𝐴𝐴𝐴𝐴𝐴 𝑥𝑥 = −𝑥𝑥 ′
𝑢𝑢 𝑥𝑥
Example: If 𝑢𝑢 𝑥𝑥 = ln 𝑥𝑥 , then 𝑅𝑅𝑅𝑅𝑅𝑅 𝑥𝑥 = 1. Therefore, log utility
implies constant relative risk aversion.
- 17 -
 Investors Choice and Utility

• Kritzman, p. 19f.
• In many applications, it is convenient to model utility as a function of
expected return and risk.

𝑢𝑢 𝜇𝜇, 𝜎𝜎 = 𝜇𝜇 − 𝐴𝐴𝜎𝜎 2

 𝑢𝑢 𝜇𝜇, 𝜎𝜎 : Utility dependent on expected return 𝜇𝜇 = 𝐸𝐸 𝑅𝑅 and risk 𝜎𝜎

 𝐴𝐴: Measure of risk aversion. 𝐴𝐴 > 0 for risk averse individuals.
The higher 𝐴𝐴, the more risk averse the individual is.
𝜕𝜕𝑢𝑢 𝜇𝜇,𝜎𝜎
 > 0: Utility increases with expected return, ceteris paribus
𝜕𝜕𝜕𝜕
𝜕𝜕𝑢𝑢 𝜇𝜇,𝜎𝜎
 < 0: Utility decreases with risk, ceteris paribus
𝜕𝜕𝜕𝜕

- 18 -
 Investors Choice and Utility

• Different choices with different risk aversions.

- 19 -
 Investors Choice and Utility

• Risk in relation to utility and return

𝑢𝑢 𝜇𝜇, 𝜎𝜎 = 𝜇𝜇 − 𝐴𝐴𝜎𝜎 2
i.e.:
𝜇𝜇−𝑢𝑢 𝜇𝜇,𝜎𝜎
𝜎𝜎 = 𝐴𝐴

- 20 -
 Investors Choice and Utility

• Risk in relation to utility and return

𝜇𝜇−𝑢𝑢 𝜇𝜇,𝜎𝜎
𝜎𝜎 = 𝐴𝐴

The graphical representation of this table gives us an indifference curve.

- 21 -
 Investors Choice and Utility

• Risk in relation to utility and return

𝜇𝜇−𝑢𝑢 𝜇𝜇,𝜎𝜎
𝜎𝜎 = 𝐴𝐴

The graphical representation of this table gives us an indifference curve.

- 22 -
Investors Choice and Utility
Indifference Curves

• The investor (with the given utility function) is indifferent between

different combinations of risk and expected return along a particular
indifference curve.
• Indifference curves that are closer to the upper left yield more utility and
thus are more desirable.
𝑢𝑢 = 𝜇𝜇 − 2𝜎𝜎 2

- 23 -
Investors Choice and Utility
Summary

Interpretation:
The higher the utility of an investment, the more attractive is the
investment to the investor.

The higher the expected return of an investment, the higher is its

utility.

The higher the risk (variance) of an investment, the lower is its

utility.

Risk averse (rational) investors like return and hate risk.

- 24 -
Investors Choice and Utility
Summary

The investor specific risk aversion is the balancing factor between

risk and return.

From a set of portfolios with different expected returns E(r) and

investment risks σ, investors will chose the most useful portfolio, i.e.
the one that maximizes their utility.
Yes, we are maximizers of
reward. But we are also
minimizers of regret.

- 25 -
Investors Choice and Utility
Summary

Within a risk-return diagram curves of constant utility U (so called

“indifference curves”) can be used to rank investments according to
their utility.
These curves are constructed in the following way: Given the utility
U the expected return E(r) is expressed as a function of risk s:

1
𝐸𝐸 𝑟𝑟 = 𝑈𝑈 + × 𝐴𝐴 × 𝜎𝜎 2
2
Introducing ½ in the equation
is just a convention;
remember slide 18!

- 26 -
Investors Choice and Utility
Summary

Interpretation:
In a risk-return diagram indifference curves have the shape of parabolas.

All investments on the indifference curve have the same utility (for the
investor).

Investments that lie to the left or above the indifference curve have higher
utility than investments below or to the right of the indifference curve.

The higher the investor’s risk aversion, the steeper the indifference curve.

- 27 -
Topic Asset Allocation

Asset Allocation

- 28 -
Asset Allocation: Complete Portfolio

Definition: The asset allocation decision is an investment decision

on how to distribute the available funds among broad investment
classes like risk-free/money market, bonds, stocks, real-estate, … .

As a result of the asset allocation process, investors should end up

with a portfolio that corresponds to their risk and return objectives.

- 29 -
Asset Allocation: Complete Portfolio

The most fundamental choice investors have to make is how to

allocate their money between the following two investment options:

I. Risky investment (P)

II. Risk free investment (F)

- 30 -
Asset Allocation: Complete Portfolio

Example:
Investment of CHF 400,000, of which CHF 100,000 are invested risk
free (money market fund) and the rest goes into two equity funds
CH-BlueChips CHF 200,000, and CH-SmallCaps CHF 100,000.

The allocation of the money follows two steps:

1. Split between risk-free and risky investment: a proportion y of
the investors total assets is invested risky and a proportion 1-y is
invested risk-free.
2. Splitting up the risky investment among all the possible risky
assets: a proportion 𝑤𝑤𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 of the risky investment is invested in
the BlueChip fund and a proportion 𝑤𝑤𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 in the SmallCap fund.
- 31 -
Asset Allocation: Complete Portfolio

Example (cont.):
Investment of CHF 400,000, of which CHF 100,000 are invested risk
free (money market fund) the rest goes into two equity funds CH-
BlueChips CHF 200,000, and CH-SmallCaps CHF 100,000.

- 32 -
Asset Allocation: Complete Portfolio

The weight y of the risky part as a percentage of the total

(“complete”) portfolio, including the risk free asset, is

𝑦𝑦 = 75% 1 − 𝑦𝑦 = 25%

Question: What happens if the investor wants to reduce the risk of

the portfolio, reducing the risky investment to 𝑦𝑦 = 50%?

- 33 -
Regrouping of investments

Example cont.: (Regrouping of investments)

The weight y of the risky part as a percentage 𝑦𝑦 = 50%

Assumption: The relative weighting within the risky part of the

investment remains constant.

- 34 -
Regrouping of investments

As long as the relative weighting within the risky fund remains

unchanged we do not have to be concerned with the individual
securities or sub-funds in the risky part of the portfolio (here: CH-
BlueChips and CH-SmallCaps). The probability distribution of the
returns of the risky part remains unchanged.

Therefore, it is sufficient to consider the risky part as one single risky

investment with fixed risk and return properties.

Analogy: view the risky investment as a fund of funds or strategy

fund with fixed proportions.
- 35 -
Regrouping of investments

However, the return distribution of the complete portfolio changes

when assets are shifted between the risk-free and risky parts.

In our further analysis in a first step we focus on complete

portfolios, i.e. portfolios that consist of only one risk-free asset (F)
and one risky asset (P).

In a second step, we will study the risk and return properties of

different combinations of two or many risky assets.

- 36 -
Expected Return and Risk of Portfolios

Having agreed on the construction of complete portfolios we now

regard the risk-return characteristics of different weightings of the
risk free and risky parts.

Example:
E( rP ) = 15%; 𝜎𝜎𝑃𝑃 = 20% ; rf = 4%

How does the risk return relationship change for different

combinations of the risk-free part (1 – y) and the risky part y?

- 37 -
Expected Return and Risk of Portfolios

Return
Leverage
ON!

Risk
- 38 -
Expected Return and Risk of Portfolios

Interpretation:
What is the risk premium of the risky part P?

𝐸𝐸 𝑟𝑟𝑃𝑃 − 𝑟𝑟𝑓𝑓 = 15% − 4% = 11%

What are the risk premium E(rC) - rf and the standard deviation σC of
a complete portfolio C? Weights are 50%-50%

𝜎𝜎𝐶𝐶 = 𝑦𝑦 × 𝜎𝜎𝑃𝑃 = 0.5 × 20% = 10%

𝐸𝐸 𝑟𝑟𝐶𝐶 − 𝑟𝑟𝑓𝑓 = 𝑦𝑦 × 𝐸𝐸 𝑟𝑟𝑃𝑃 − 𝑟𝑟𝑓𝑓 = 0.5 × 11% = 5.5%
Also: (50% * 15% + 50% * 4%) – 4% = (7.5% + 2.0%) – 4% = 9.5% - 4%

- 39 -
Expected Return and Risk of Portfolios

Consider the ratio S between risk and risk premium:

100% risky portfolio

𝐸𝐸 𝑟𝑟𝑃𝑃 − 𝑟𝑟𝑓𝑓 15% − 4%

S= = = 0.55
𝜎𝜎𝑃𝑃 20%

This Ratio S describes the reward (return) you get for taking one
(additional) unit of risk.

Definition: The slope S of the Capital Allocation Line (CAL) is a

parameter which describes the reward-to-variability ratio. It is also
known under the name Sharpe-Ratio.
- 40 -
Capital Allocation Line (CAL)

Comments:
1. Risk 𝜎𝜎𝐶𝐶 and risk-premium 𝐸𝐸 𝑟𝑟𝐶𝐶 − 𝑟𝑟𝑓𝑓 of a complete portfolio C
are proportional to the share y of the risky investment.
2. The reward-to-variability expresses, what excess return an
investor can expect for one additional unit of risk.
3. The reward-to-variability ratio remains constant for all possible
complete portfolios, i.e. for all portfolios on the CAL.
4. Two equivalent formulas for the return of the complete portfolio
Mathematically, this is a
𝐸𝐸 𝑟𝑟𝐶𝐶 = 𝑟𝑟𝑓𝑓 + 𝑦𝑦 × 𝐸𝐸 𝑟𝑟𝑃𝑃 − 𝑟𝑟𝑓𝑓 straight line…

𝐸𝐸 𝑟𝑟𝐶𝐶 = 𝑦𝑦 × 𝐸𝐸 𝑟𝑟𝑃𝑃 + 1 − 𝑦𝑦 × 𝑟𝑟𝑓𝑓

The first one is called the equation of the capital allocation line.

- 41 -
Capital Allocation Line (CAL)

Can we construct complete portfolios with a higher risk-premium than the risky
portfolio P? → increase the share y of the risky investment above 1 !
Leverage!

If y>1 than the risk-free investment 1-y is negative, which means borrowing
money (taking a loan). The resulting complete portfolio is called a leveraged
portfolio.

Example: Available capital CHF 100,000

Additional capital (loan) CHF 50,000
Total risky investment CHF 150,000

150,000
Share of risky investment: 𝑦𝑦 = = 1.5
100,000

Risk-free investment (loan!): 1 − 𝑦𝑦 = 1 − 1.5 = −0.5

Complete portfolio return 𝐸𝐸 𝑟𝑟𝐶𝐶 = 𝑦𝑦 × 𝐸𝐸 𝑟𝑟𝑃𝑃 + 1 − 𝑦𝑦 × 𝑟𝑟𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏
- 42 -
Capital Allocation Line (CAL)

Example (cont.):

Complete portfolio:
Expected Return: 𝐸𝐸 𝑟𝑟𝐶𝐶 = 𝑦𝑦 × 𝐸𝐸 𝑟𝑟𝑃𝑃 + 1 − 𝑦𝑦 × 𝑟𝑟𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏
𝐸𝐸 𝑟𝑟𝐶𝐶 = 1.5 × 15% + (−0.5) × 4% = 22.5% − 2% = 20.5%

Risk: 𝜎𝜎𝑐𝑐 = 𝑦𝑦 × 𝜎𝜎𝑃𝑃

𝜎𝜎𝑐𝑐 = 1.5 × 20% = 30%

Slope S of the Capital Allocation Line:

𝐸𝐸 𝑟𝑟𝐶𝐶 −𝑟𝑟𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 20.5%−4%

S= 𝜎𝜎𝐶𝐶
= = 0.55
30%

Attention: S remained at 0.55. (because 𝑟𝑟𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝑟𝑟𝑓𝑓 )

- 43 -
Capital Allocation Line (CAL)

Example: Excel

Return

Leverage
ON!

Risk
- 44 -
Capital Allocation Line (CAL)

More realistic scenario: What happens if we allow for a borrowing rate

that is higher than the risk-free rate?
Now rf = 4% as before, but
rB = 8% > rf
Which of the following three numbers changes?
𝐸𝐸 𝑟𝑟𝐶𝐶 = 1.5 × 15% + (−0.5) × 8% = 22.5% − 4% = 18.5%
𝜎𝜎𝑐𝑐 = 1.5 × 20% = 30%
Slope of the CAL for leveraged portfolios is now:
𝐸𝐸 𝑟𝑟𝐶𝐶 − 𝑟𝑟𝐵𝐵 18.5% − 8%
Sleveraged = = = 0.35
𝜎𝜎𝐶𝐶 30%
Due to higher borrowing costs, any additional unit of risk beyond the risky
portfolio P is rewarded by only 0.35% additional risk premium!
- 45 -
Capital Allocation Line (CAL)

Example: Excel

Return

Leverage
ON!

Risk
- 46 -
Capital Allocation Line (CAL)

Graphic Representation:
Slope of the CAL to the left of P («normal» portfolios): S=0.55
Slope of the CAL to the right of P («leveraged» portfolios): SL=0.35

- 47 -
Capital Allocation Line (CAL)

Why is the Slope

lower?

Slope S

- 48 -
Topic Asset Combination

Asset Combination
and Diversification

- 49 -
Capital Allocation Line (CAL)

Question: How do investors choose among complete portfolios on

two different CALs?

CAL 1

CAL 2

Remark: Every investor will only choose portfolios on the CAL with
the higher Sharpe-ratio.
- 50 -
Capital Allocation Line (CAL)

Investing in risky asset naturally comes with the risk of incurring

losses.

One of the most important insights in modern finance is:

Combining several risky assets within a portfolio effectively stabilizes

the return by lowering the portfolio risk.

This follows the well-known dictum:

“Don’t put all your eggs in one basket”.

- 51 -
Diversification and Portfolio Risk

Example: Take two arbitrary assets and construct a portfolio

consisting of these two assets, each with a weighting 𝑤𝑤 between 0%
and 100%.
Question: What is the risk-return characteristic of the resulting
portfolio?
Return/Risk of a combined Portfolio

Expected Return

Risk
- 52 -
Diversification
Correlation vs. Cosinus theorem

• The Law of Cosines (also called the Cosine Rule) is very useful for
solving triangles:

• Law of Cosines c 2 = a 2 + b 2 − 2 ⋅ a ⋅ b ⋅ cos(γ )

Note:

- 53 -
Diversification
Correlation vs. Cosinus theorem

• Two positions a (Nestlé) and b (SMI-ETF) with volatilities σa, σb and

correlation ρa,b
• We combine the two positions to form a portfolio c = a + b

σ2c = σ2a + σb2 + 2 ⋅ σa ⋅ σb ⋅ ρa,b

• We can interpret this geometrically

Equivalent to
vector
addition! σc cos(γ ) = −ρa,b
σb
γ cos(180° − γ ) = ρa,b
σa
- 54 -
Diversification
Correlation vs. Cosinus theorem
σa σb
γ = 180o
ρ=1

Correlation tells us
σc “how much of b lies
σc
in the direction of a”
ρ = 0.5
σb ρ = 0 σb
γ = 120 o
γ = 90o
σa
σa
σc
σb
ρ = -0.5 γ = 0o
γ = 60o σc σb ρ = -1
σa
- 55 -
Topic Asset Combination

Diversification
and Portfolio Risk

- 56 -
Diversification

• Portfolio with 100 assets

• The total risk of the portfolio is given as
σP2 = ∑ w i2σi2 + ∑ σij w i w j
i i≠ j

• How may variance terms?

• How many covariance terms?
• What happens if we add an other instrument to the portfolio?
𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑: Variance,
covariance & correlation…

- 57 -
Diversification

• Assume that we have n equally weighted assets in a portfolio

• For the Portfolio variance we get
1 2 1
σ = ∑ 2 σ i + ∑ 2 σ ij
2 For a portfolio with many equally
p weighted assets:
i n i≠ j n - the variances of the assets
diversify away but
 ∑ σ i2   ∑ σ ij  - the risk caused by the co-
movements of the assets can
1  + n − 1  i≠ j 
=  i never be diversified away
n n  n  n(n − 1) 
   
 
𝐍𝐍𝐍𝐍𝐍𝐍𝐍𝐍: For very large n, 1/n
1 2  1 will be very small aka 0
= σ avg + 1 − σ avg
n var_avg  n  covar_avg

- 58 -
Diversification
The mathematical way…

• The Covariance Matrix contains the information of how assets

move and interact
Asset 1 Asset 2 Asset N
How do How do

Asset 1
How does

Asset 1
Asset 2 & Asset N &
2
Asset 1 Asset 1 co- … Asset 1
σ σ1,2 …
 σ1,N  move?
move? co-move?
 1
2
 How do
 σ2 ,1 σ2   …
How do

…
How does

Asset 2

Asset 2
Asset 1 &
V = Asset N &

σN−1,N 
Asset 2 Asset 2
Asset 2 co-
   move?
…

co-move? move?
σ … σN,N−1 σN2 

…
 N,1 

…
How do How do

Asset N
How does

Asset N
Asset 1 & Asset 2 &
Asset N co- Asset N co- … Asset N
move?
move? move?
Asset 1 Asset 2 Asset N

- 59 -
Diversification
The mathematical way…

• We will only use matrix notation as an easy way to write and

interpret formulas
• We do not have to know the whole theory behind linear
objects (linear algebra)

N elements per column

3 elements per column

Row 1 R1C1 R1C2 R1C3 R1C4
Column 1
Column 2
Column 3
Column 4
Row 2 R2C1 R2C2 R2C3 R2C4

Row 3 R3C1 R3C2 R3C3 R3C4

N elements per row

4 elements per row

• If Row 2 = (a1, a2,…, aN) and Column 3 = (b1, b2,…, bN), then
R2C3 = (a1b1 + a2b2 +… + aNbN)
- 60 -
Diversification
The mathematical way…
𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑𝐑: Variance,
covariance &
• Volatility • Expected return over cash
correlation…
– Asset A: 20% – Asset A: 12%
– Asset B: 25% – Asset B: 15%
• Correlation between A and B: 0.5

 4% 2 .5 %   1 0 .5   0 .4   0.12 
V =  C=  w =  μ= 
 2.5% 6.25%   0 .5 1   0 .6   0.15 

 4% 2 .5 %   0 .4 
σ = w ' ⋅ V ⋅ w = (0.4 0.6 ) ⋅ 
2
P ⋅ =
 2 . 5 % 6 . 25 %   0 . 6 
 0 .4 
= (3.1% 4.75% ) ⋅   = 4.09%
 0 . 6 
σP = 20.22%
 0.12 
µP = w ' ⋅ μ = (0.4 0.6 ) ⋅   = 13.8%
 0.15 

- 61 -
Diversification and Portfolio Risk

Remark:

• Diversification benefits (risk reduction) exists as long as the

returns of the two assets do not always behave in exactly the
same way, i.e. if the correlation 𝜌𝜌 of their returns is 𝜌𝜌 < +1!

• Only if the two assets behave always in exactly the same way (i.e.
their correlation is 𝜌𝜌 = +1), then the effect of diversification
ceases to exist.

- 62 -
Topic Asset Combination

Insertion:
Splitting the risk

- 63 -
Diversification and Portfolio Risk

What happens if we continue this diversification strategy?

We construct portfolios with more and more assets:

n = 10 100 1000 10000 100000 …

(n is the number of assets within the portfolio)

Question:
How will portfolio risk or portfolio variance develop, as the number
of assets in the portfolio grows?
- 64 -
Diversification and Portfolio Risk

Systematic Risk

- 65 -
Diversification and Portfolio Risk

Interpretation:

The variance (risk) of a portfolio decreases (quite rapidly) to the

level of the so called systematic risk if more and more assets are
included in the portfolio.

Generally, “good” diversification (i.e. a significant reduction of

specific risk) can already be achieved with about 15 to 20 assets.

𝐍𝐍𝐍𝐍𝐍𝐍𝐍𝐍: E.g. CTAs may

have up to 200

- 66 -
Splitting the Risk

Systematic risk:

A risk that is carried by an entire class of assets or to the entire

global economy. It is impossible to reduce systematic risk for the
global economy by diversification. Systematic risk is the risk that
remains in fully diversified portfolio (e.g. risk of a broad stock index).

But: One may mitigate some of the systematic risk by buying

securities of different asset classes and/or by buying in different
markets/countries.

- 67 -
Splitting the Risk

Question:
How can we understand this risk reduction and the remaining level
of systematic risk?

Following a top-down-approach risk emerges on several levels. For

instance:
Markets: GDP growth, interest rates, geopolitical events …
Sector: commodity prices (e.g. oil); regulatory changes …
Company: CEO resigns or new CEO; launch of new product;
internal fraud …

- 68 -
Splitting the Risk

The risk is not in the listed factors per se, but in the uncertainty of
their future development and – most importantly – in the fact that
we do not know the influence of this development on future
returns.

If shares of more and more companies and sectors are included in

the portfolio, the effects of company and sector specific risk factors
compensate (diversify away) and only the effect of the systematic
or market risk factors remain in the portfolio returns.

- 69 -
Splitting the Risk

The risk of a single asset or a portfolio can be split into two parts:

Risk

Systematic Unsystematic

Non-diversiviable diversiviable

Unique, company-specific or
Market risk
idiosyncratic risk

- 70 -
Splitting the Risk

Systematic risk, to a large extent, results from macro-economic

factors and influences all assets, while title or company-specific risk
is of concern only to the company in question.

Through a skillful choice of assets and markets (⇒ national and

international diversification) title-specific risk can, to a large extent,
be eliminated. What remains in the portfolio is systematic or market
risk.

- 71 -
Splitting the Risk

𝐍𝐍𝐍𝐍𝐍𝐍𝐍𝐍: Nowadays, more and

Modern PF-Theory (MPT) states: more Alternative Risk
Premiums are identified!

Only systematic risk is rewarded by a sustainable risk premium.

Systematic risk reflects the reaction of individual assets and

portfolios to fluctuations in the market.

Remark: The beta factor (𝛽𝛽 factor) describes how sensitively

individual assets and/or portfolios react to fluctuations in the
market and/or macro-economic factors.

- 72 -