INDIAN INSTITUTE OF TECHNOLOGY

KANPUR

Lesson: Introduction to risk and return

Advanced Algorithmic Trading and Portfolio

Management
Introduction
In this lesson we will cover the following topics:

• Basics of risk-return framework

• Measures of risk

• Diversification of risk

• Computing portfolio risk

• Impact of individual securities on portfolio riskSummary and

Concluding remarks
Basics of Risk-return framework

• Consider three instruments: T-Bills, Government Bonds, and

common stock

• T-Bills are short maturity instrument with almost no risk of default

• Bond is a rather long-term instrument and fluctuates with interest

rates

• Common stocks are infinite maturity instruments

Basics of Risk-return framework
T-Bill Returns=4% Bond Returns =5.5% Common Stock Returns =11.1%
Basics of Risk-return framework
• Notice the difference between the returns on T-Bills and common stocks: 11.1-4=7.1%

• This additional return can also be said to be the risk-premium received by investors

• On a given year T-Bill rate was 0.2% and you are asked to estimate the expected return
on common

stocks. A reasonable estimate would be obtained by adding this 7.1% to obtain the total
return of 7.30%

• However, this assumes that there is a stable risk premium on the common stock portfolio,

that is, future risk premium can be measured by the average past risk premium

• But (a) Economic and financial conditions change overtime; (b) Risk perceptions change;

(c) Investors' risk tolerance and return expectations also change over time
Basics of Risk-return framework
• Consider a stock with $12 dividend expected by the end of
the year

• Investors are expecting a 10% return on this stock

= $400; Dividend yield =

1
P𝑉 =𝑟−𝑔
𝐷𝐼𝑉1
2
=
•
0.10−0.07 12/400=3%.

• If dividend yield changes to 2%, and investors demand an expected return =

2%+7%=9%

𝑃𝑉 =
1
2
0.09−0.0
= $600
•
7

𝑃
•
dividends: 𝑟 = 𝐷𝐼𝑉1 + 𝑔
Expected returns on the stock reflect the dividend yields and the growth rate of
0

• Risk-premium= 𝑟 − 𝑟𝑓; this risk premium can change

overtime

• Often dividend yield is a good indicator of risk-

premium
Measures of risk
• A very prominent statistical measure of risk is variance (or standard
deviation)
• 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑟𝑡 = 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑
𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑟𝑡 − 𝑟
ҧ

• Where 𝑟𝑡 is the actual return and 𝑟ҧ2is the

expected returns
• 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑆 = 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒
𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 𝐷 (𝑟𝑡)
• Standard deviation is often denoted by the symbol 𝜎 and
variance by 𝜎2
Measures of risk
• Let us understand this concept with a small coin toss game

• The following probabilities are observed

 (a) H+H: Gain 40%; (b) H+T: Gain 10%;

 (c) T+H: Gain 10%; (d) T+T: lose 20%

• Thus, there is a 25% chance that your return will be 40%, 50% chance that

your return will be 10%, and 25% chance that you will lose 20%

• Expected return : 𝑟ҧ=0.25*40%+0.5*10%+0.25*(-20%)= 10%.

Measures of risk
• Now let us compute the variance and standard deviation of
these returns

Mean Deviation (𝐫𝐭 − 𝐫ҧ)

𝐫𝐭 − 𝐫ҧ 𝟐
Returns (%) Mean Square deviation Probability Probability squared deviation

40 30 900 0.25 225

10 0 0 0.50 0
-20 -30 900 0.25 225
Total 450

• Variance= 225+225=450 and Standard Deviation (𝜎)= (450)=21%

• An event is considered to be risky if there are many possibilities of outcomes associated with it

• As these possibilities increase, i.e., the spread of possible outcomes increases, the event is said to have
become riskier

• Standard deviation or variance is a summary measure of these possibilities, that is spread in the possible
outcome
Measures of risk
• The risk of an asset can be completely expressed, by writing all the possible outcomes and the possible
payoffs associated

with each of the outcome

• If the outcome was certain, i.e., no risk, then the standard deviation would have been zero

• One of the challenges in performing such computations is the estimation of probability associated with each
outcome

• One way to go about this is to observe past variability

• Portfolio
For example, consider the historical Standard Deviation
volatilities (𝛔) different kinds ofVariance
of three (𝛔𝟐)
securities
Treasury Bills 2.8 7.7
Government Bonds 8.3 69.3
Common Stocks 20.2 406.4

• It appears that T-Bills are the least variable and common stocks are the
most variable
Diversification of risk
• One can compute the measure of variability for individual
securities as well as the portfolio of

securities

• The standard deviation of selected U.S. Common stocks (2004-08) such as Amazon

(50.9%), Ford (47.2%), Newmont (36.1%), Dell (30.9%), and Starbucks (30.3%) was

much less than the standard deviation of market portfolio, i.e., 13% during this

period

• It is well known that individual stocks are more volatile than the market indices

• The variability of market doesn’t reflect or is same as the variability of individual stock
components
Diversification of risk
Diversification of risk
SD of Dell and Starbucks is approximately 30%

SD of portfolio = 20%
Diversification of risk
Computing portfolio risk
• We now know that diversification reduces the risk of a
portfolio

• Consider a portfolio comprising stocks A (60%) and B (40%)

• A has expected returns of 3.1% and B has expected returns

of 9.5%

• 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑟𝑒𝑡𝑢𝑟𝑛 = 0.6 ∗ 3.1 + 0.40 ∗ 9.5 =

5.7%

• Standard deviation of A is observed as 15.8% for A and

23.7% for B

• Standard deviation of this portfolio: 0.6*15.8%

 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 = 𝑥 2 𝜎 2 + 𝑥 2 𝜎 2 +
Computing portfolio risk 2(𝑥1𝑥2𝜌12𝜎1𝜎2)
1 1
𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 =
2
2
𝜎1
𝜌12𝜎1𝜎2 2

Correlation coefficient (𝜌12)

= 𝜌12𝜎1𝜎2
Computing portfolio risk
• Let us fill the above box with some numbers; Assume a correlation
coefficient of 1
𝑥2 𝜎 = 0.62 ∗ 15.82 𝑥1𝑥2𝜎1𝜎2 = 0.6 ∗ 0.4 ∗ 1 ∗ 15.8
Stock A Stock B
2
1 1
∗ 23.7
Stock A

𝑥1𝑥2𝜎1𝜎2 = 0.6 ∗ 0.4 ∗ 1 ∗ 15.8 ∗ 23.7 𝑥22 𝜎22 = 0.42 ∗ 23.72

• Portfolio variance= 0.62 ∗ 15.82 + 2 ∗ 0.6 ∗ 0.4 ∗ 1 ∗ 15.8 ∗ 23.7 + 0.42

Stock B

∗ 23.72 = 359.5
359.5 =
19%
• The standard
deviation is
• Let us now assume a correlation coefficient of 𝜌12 = 0.18

• Portfolio variance= 0.62 ∗ 15.82 + 2 ∗ 0.6 ∗ 0.4 ∗ 0.18 ∗ 15.8 ∗ 23.7 + 0.42
∗ 23.72 = 212.1
212.1 =
14.6%
• The standard
deviation is
Computing portfolio risk
correlation 𝜌12 = −1
• Let us consider a very hypothetical case of extreme negative

• Portfolio variance= 0.62 ∗ 15.82 + 2 ∗ 0.6 ∗ 0.4 ∗ (−1) ∗ 15.8 ∗ 23.7 +

0.42 ∗ 23.72 = 0!

• However, perfect negative correlations do not exist in real markets

Computing portfolio risk
• Variances in diagonal boxes (𝑥 2 𝜎 2 )

• Covariance terms in off-diagnol (𝑥𝑖𝑥𝑗𝜎𝑖𝑗)

• Let us consider a case of N securities and

equal
investment in all the
securities ( 1 )
�
�

• Portfolio variance can be computed in the

form of

two components. That is,

variance component and

Computing portfolio risk
∗ (𝐴𝑣𝑒𝑟𝑎𝑔𝑒
1

written as 𝑁 ∗ 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒)
• There will be N variance terms; then portfolio variance can be simply
𝑁2

Remember 𝑤
1 2∗ 𝑤∗ 𝜎 . 𝐻𝑒𝑟𝑒
2
1 𝑤 =𝑤 = 1
; 𝑎𝑛𝑑 𝜎 = 𝐴𝑣
𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 =𝜎
•
2 𝑔
�

Also, 𝑁 2 − 𝑁 covariance terms where average covariance term

�
•
=𝜎𝐶𝑜𝑣−𝐴𝑣𝑔
The sum of covariance terms is 𝑁 2 − ∗
1
𝐶𝑜𝑣−𝐴𝑣
𝑁 ∗ 𝜎
•
𝑁2 𝑔

• 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 ∗ 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑉𝑎𝑟𝑖𝑛𝑎𝑐𝑒 ∗ (𝐴𝑣𝑒𝑟𝑎𝑔𝑒

1 1

=𝑁∗ 𝑁2
+ 𝑁2 − 𝑁 ∗ 𝑁2
𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒)

∗ 𝐴𝑣𝑒𝑟𝑎𝑔𝑒
1

risk term, 𝑁 ∗ 𝑉𝑎𝑟𝑖𝑛𝑎𝑐𝑒 ,

• As the number of securities, N, in the portfolio increase, the specific-
𝑁2

approaches to a value of
zero
Computing portfolio risk
• Thus, the overall portfolio variance approaches the average covariance term

• This is also often referred to as portfolio diversification

• Thus, if these securities have very low correlation, then one can obtain a portfolio

with very low risk

• That is, just by increasing the number of securities in a portfolio,one can eliminate

the idiosyncratic (specific or diversifiable risk)

• The remaining risk is often called market risk or non-diversifiable risk

• That is why, this market risk (or average covariance or non-diversifiable risk) is

what constitutes the bedrock of risk, that is risk that is there after eliminating all the
Impact of individual securities on portfolio risk
• Investors usually add many securities in their portfolio to diversify the stock-specific
idiosyncratic risk

• It is not the risk of a security held individually but in a portfolio that is important

• To measure the impact of a security to the risk of portfolio, one needs to measure the

market risk component of the security

• The market risk of a security is measured through its beta

• Stocks with beta of more than 1.0 tend to amplify the movements of market

• Stocks with beta between 0 to 1.0 tend to move in the same direction as market, but

are considered less sensitive

• The market portfolio has a beta of 1.0 and reflects the average movement of all the
Impact of individual securities on portfolio risk
• Consider a stock A with beta of 1.41 over a given time-

horizon

• This means that, on average, when market rises by 1%,

stock A will rise by 1.41%

• The stock would also have some stock-specific risk

• When a stock is added to a well-diversified portfolio, the movements on account of

idiosyncratic

factors are expected to cancel each other out

• Therefore, for this portfolio what matters is only these systematic market related
Impact of individual securities on portfolio risk
• Stocks like Stock A with high beta will have steep straight
curve

• Stocks with small beta (e.g., beta=0.3), the

straight line plot will be less steep

• A stock with high beta may also have less

idiosyncratic risk and a stock with low beta
may also have high idiosyncratic risk
• For example, a stock of gold-mining firm may have
low beta and a very high idiosyncratic stock

specific risk

• When added to a well-diversified portfolio, the

Impact of individual securities on portfolio risk
• So, let us now answer this question how security betas
affect the portfolio risk

• Market risk accounts for most of the risk of a

well- diversified portfolio

• Beta of an individual security measures

its sensitivity to
market movements

• Examine the figure shown here: the standard deviation (total risk) of the portfolio

depends on the number of securities in the portfolio

• As the number of securities increase in the portfolio, more diversification is

achieved
Impact of individual securities on portfolio risk
• With addition of more and more securities, the
specific risk declines until all the stock specific
risk is eliminated and only market risk remains

• Market risk depends on the average beta of the

securities,
i.e., the portfolio beta

• If one selects a fairly large number of securities

from a market, you diversify all the idiosyncratic
risk
• Thus, you get the market portfolio with beta= 1.0

• If the market portfolio has a standard deviation of 20%, then this portfolio is
Impact of individual securities on portfolio risk
Impact of individual securities on portfolio risk
• Beta of a stock ‘i' can be computed using the following formula. 𝛽𝑚𝑖 = 𝜎𝑖𝑚 /𝜎2 .
Here 𝜎𝑖𝑚 is the
𝑚
covariance between the stock returns and market returns. 𝜎2 is the
variance of the returns on
the
market.
1 2 3 4 5 6 7
Month Market return Deviation in Market Squared Market Stock A Deviation in Stock Deviation Product
(%) Returns Deviation A returns (3*6)
1 -8 -10 100 -11 -13 130
2 4 2 4 8 6 12
3 12 10 100 19 17 170
4 -6 -8 64 -13 -15 120
5 2 0 0 3 1 0
6 8 6 36 6 4 24

Variance= σ 𝑚 = = 50.67
304
Avg.= 2 Sum=304 Avg.= 2 Sum=456
2
6

= 76
456
Co-variance=σ𝑖𝑚= 6

Beta= 𝛽𝑖 = =50.67= 1.5

σ𝑖𝑚 76
σ𝑚
2
Impact of individual securities on portfolio risk
• Can we say that a diversified firm is more attractive to investors than an undiversified
firm

• If diversification is a good objective for a firm to pursue then each new project’s
contribution to firm diversification should also add value to the firm

• This seems to be not consistent with what we have studied about present values

• Investors can diversify for themselves more easily than firms

• If investors can diversify on their own, they would not be paying anything extra
to firm for this diversification

• The present value of any number of assets is equal to the present value of their
parts. That is, PV (ABC)=PV(A)+PV(B)+PV(C): Value Additivity
Summary and Concluding remarks

• Returns to investor vary depending upon the risk borne by them

• Very safe instruments such as treasury securities provide the lowest returns

• Equity securities are considered to be more riskier asset class and offer higher
expected returns

• Accordingly, the discount rates applied to a safe project versus risky project will also
differ

• Risk of a security means that there are many possible return outcomes for that
security

• The total risk of a stock has two components: Stock-specific risk and Systematic (or
Summary and Concluding remarks
• Investors eliminate a sizable portion of their specific (or diversifiable) risk, simply by adding
more securities

to their portfolio

• A well diversified portfolio is only exposed to market risk

• A security’s contribution to a well diversified portfolio measured as the sensitivity of the

security to market movements, i.e., beta (β)

• A stock with high beta is more sensitive to market movements and vice-versa

• Investors can diversify on their personal account, they do not want firms to pursue the

diversification objective
Thanks!
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Lesson: Portfolio Theory and Asset Pricing Models

Advanced Algorithmic Trading and Portfolio

Management
Introduction
In this lesson we will cover the following
topics:
•Investment performance and return
distribution
•Combining stocks with portfolios
•Introduction to CAPM
•Validity of CAPM
•Alternative theories of asset pricing
•Summary and concluding remarks
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Investment Performance and Return

Distribution
Investment Performance and Return Distribution

Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Investment Performance and Return Distribution

• Compare investments A and B. These

investments offer an expected returns
of 10%. But A has much wider
spread of possible outcomes (SD
of A is 15% and that of B is 7.5%).
• Compare investments B and C. Both
of them have the same standard
deviation. However, the expected
returns from B (10%) and C (20%)
are different.

Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Combining Stocks with Portfolios:

Part 1
Combining Stocks with Portfolios
• Consider a scenario where you are examining stocks A and B
as potential investments.
• Stock A offers 3.1% expected returns and Stock B offers 9.5%
expected returns.
• Stock A has a standard deviation of 15.8% and stock B has
a standard deviation of 23.7%.
• You can invest in a combination of these stocks.
• If you invest 60% in stock A and 40% in stock B then the
expected return from this
portfolio, is 0.60*3.1% + 0.40*9.5% = 5.66%.
• The same can not be said about the risk of the portfolio.
Combining Stocks with Portfolios
•The risk of a portfolio, that is standard deviation (SD), is
less than the simple weighted average of individual
Variance
•stock SDs.= x2 σ2 + x2 σ2 + 2 ∗ x x σ σ = 0.602 ∗
1 1 2 1 2 1 2
23.72 + 2 ∗220.60 ∗
15.8 + 0.4 ∗
2 0.40 ∗ 0.18 ∗ 15.8 ∗ 23.7 =
212.1;
Standard Deviation = Sqrt 212.1 =
14.6%
•The lower amount of SD reflects the diversification
aspect, assuming a correlation of 0.18.
Combining Stocks with Portfolios
• The blue curve line shows all the
possible expected risk and return
combinations of these two stocks
that one can achieve.
• A risk averse investor would hold
A:B (50:50
or 60:40)
• A less risk-averse investor would
invest most of their wealth in B.

• The brown line connecting A and B

represents all portfolio combinations
withMyers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
Brealey,
editions. Chapter 8
Combining Stocks with Portfolios
• In practice, you invest in many
socks, by examining their
historical risk- return related
properties.
• For example, consider a
portfolio of ten securities
plotted here using risk- return
data.
• The shaded green region
shows the possible
combinations of expected
return and standard deviation
by investing in a mixture of
Combining Stocks with Portfolios
• Where would you want to be
in that shaded region?
• You would want to go up, that
is, increase the expected
returns. You would also
want to go left, that is, to
reduce risk.
• As you move up and left, you end
up at the solid dark brown line.
• The portfolio on this solid dark
outer surface is often referred to
as
Brealey,an efficient
Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Combining Stocks with Portfolios
• For a given level of risk, these
portfolios offer the highest
return. And for a given
level of return, these
portfolios offer the lowest
amount of risk.
• Three such portfolios (A, B,
and C) are shown in the
figure here.

• You want to deploy the

investor’s funds to generate
maximum expected
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Combining Stocks with Portfolios:

Part 2
Combining Stocks with Portfolios
• Now we introduce the
possibility of lending and
borrowing at a risk-free rate
of interest (rf).
• Is this possibility a practical
scenario?
• Acombination of rf and any
efficient portfolio (e.g., S) can
offer various risk- return
possibilities on the line rf-S.
• Investing in rf and S leads a
portfolio
Brealey, on
Myers and thePrinciples
Allen; line segment
of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Combining Stocks with Portfolios
• Suppose that portfolio S has an expected return of 15% and a standard
deviation of 16%.
• For risk-free instrument rf = 5% and risk = 0.
• If you decide to invest 50% in S and 50% in rf, the expected return
and risk as computed here.
• 𝑟 =1 ∗ 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑆 + ∗
1
2
𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡
2 𝑅𝑎𝑡𝑒 𝑜𝑛 𝑅𝑖𝑠𝑘 − 𝑓𝑟𝑒𝑒 𝑥2 𝜎2 + 𝑥2 𝜎2 +
2𝑥1𝑥2𝜌12𝜎1𝜎2
=
risk: 𝑆𝐷 =
• The formula for computation of [Here
1 1 2 2
10%

•𝜎
,
𝜎2== 0]∗ 𝑆𝐷 𝑜𝑓 𝑆 =
2
1
0.5*15% = 8%
Combining Stocks with Portfolios
•Consider another scenario where you borrow at the
risk-free rate an amount equal to 100% of your initial
wealth.
•You invest your initial 100% wealth along with these
•borrowings 𝑟 =
2 ∗ S.
in Portfolio
Expected returns: 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛
That is double the𝑜𝑛 𝑆
amount of

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒ሻ = 25% − ሺ1

your initial wealth.

•Risk 𝜎 = 2 ∗ 𝑆𝐷 𝑜𝑓∗𝑆
= 32%
Combining Stocks with Portfolios
• On the efficient region, you can
always find a portfolio S that is the
best efficient portfolio.
• How to find this portfolio?
• The steepest line (from rf) on the
curve representing efficient
portfolios: tangent line
• This tangent line has the highest ratio of risk-premium to standard
𝑟−𝑟
deviation: Sharpe Ratio
• 𝑆ℎ𝑎𝑟𝑝𝑒
𝑅𝑖𝑠𝑘−𝑃𝑟𝑒𝑚𝑖
= �
𝑓
𝑢𝑚
𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑
𝑟𝑎𝑡𝑖𝑜 = 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜 �

Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Combining Stocks with Portfolios
•In a competitive market,
it is extremely difficult
to find undervalued
securities.
•Professional investors
often investment
in benchmark indices
(e.g., S&P 500).
•This is often referred to as
the passive strategy of
Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
investment.
editions. Chapter 8
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Introduction to CAPM
Introduction to CAPM
•We have previously examined the returns on different
instruments.
•T-Bills have a beta = 0, and the market portfolio has a
beta = 1.
•Difference between market risk (rm) and risk-free rate
(rf) is often referred to as market risk premium.
•Using these benchmarks, we can determine the risk-
premium for instruments for which beta is neither 0
nor 1.
Introduction to CAPM
• In 1960s, three economists,
Sharpe, Lintner, and
Treynor came-up with this
model called Capital Asset
Pricing Model (CAPM) that
provides an extremely
simple and easy to use
solution for the asset pricing
problem.
• In a competitive economy, the
risk- premium is directly
proportional to beta.
Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Introduction to CAPM
• The risk-premium on an investment with beta of 0.5 should be half of
that available on the market.
• The expected risk-premium on an investment with beta of 2 is twice
the risk-premium
expected on the market.
• The resulting relationship is shown here: 𝑟 − 𝑟𝑓 = 𝛽 ∗ ሺ𝑟𝑚 − 𝑟𝑓ሻ
• Consider two stocks with beta of 0.30 (Stock A) and 2.16 (Stock B).

ሺ𝑟𝑚 − 𝑟𝑓ሻ
You also observe that the market is offering a current risk-premium of
𝑟𝐴 = 𝑟𝑓 𝑟𝑚 − = 0.20% + 0.30 ∗ 7% =
+𝛽∗ the current𝑟𝑓treasury
2.30%
• 7%

• 𝑟𝐵 = 𝑟𝑓 𝑟𝑚 − = 0.20% + 2.16 ∗ 7% =
and bill rate is 0.2%.

+𝛽∗ 𝑟𝑓 15.32%
Introduction to CAPM
• CAPM can also be employed to estimate discount rates for risky
projects and companies.
• To estimate discount rates, different risk factors, appropriate
benchmark for risk-free
rates needs to be estimated.
• The following principles are sacrosanct:
• Investors like higher expected returns and low risk.
• If the investors can lend and borrow at risk-free rate of interest,
then one portfolio
is better than all the other portfolios.
• This best efficient portfolio depends on (a) expected returns,
(b) standard deviation, and (c) correlations across securities.
• In a well-diversified portfolio, only systematic risk matters.
Introduction to CAPM
•If stocks A and B
(overvalued) do not fall
on this line, then you
will not buy them.
•Given the less demand
and excess supply, the
prices of A and B will
fall until the expected
returns lie on SML.
•The same logic applies
Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
to Chapter
editions. undervalued
8
Introduction to CAPM
• Investors can hold a
combination of market
portfolio M and risk-free rate
rf, to obtain an expected
return 𝑅ത =
𝑟𝑓 + 𝛽 𝑟𝑚 − 𝑟𝑓
• In well-functioning liquid and
efficient markets, nobody
will hold a stock that offers
anything less.
• Equilibrium
Brealey, is obtained
Myers and Allen; from Finance.
Principles of Corporate the arbitrage mechanism, which
10th, 11th, or 12th
editions. Chapter 8
drives prices towards efficient values, that is, towards this
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Validity of CAPM
Validity of CAPM
• Any economic model aims to provide a simple view of actual
and real-world scenarios.
• There is a trade-off that the real thing may be far-away from the
model if the model is too simple.
• Otherwise, the complexity has to be increased to make it closer
to the real thing.
• Investors are rational, risk-averse individuals that require
extra-return for taking on additional risk.
• Investors do not worry about those risks that can be
diversified.
• The power of CAPM lies in its extreme simplicity, and it also
has some pitfalls.
Validity of CAPM
• Ten investors portfolio
returns are plotted.
• Investor 1 has a portfolio of
mostly small stocks and
Investor 10 has a portfolio of
large-cap stocks.
• One can obtain by combining
Investor 1 (long) and investor
(10) to generate a zero-
risk portfolio that offers excess
abnormal
Brealey, Myers and returns.
Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Validity of CAPM
• The red line shows the cumulative
difference between small and
large cam firms.
• The green line shows the
cumulative difference between
high book to value (Value
stocks) minus low book to value
stocks (Growth stocks).
• The figure does not fit well with
CAPM postulations: that is
beta is the only factor
causing returns to differ across
instruments.
Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
editions. Chapter 8
Validity of CAPM
•Value stocks are underpriced cheap stocks. They
may be underpriced at current P/E ratios for
different reasons.
•Growth stocks are not cheap stocks at current
P/E levels.
•The returns on value stocks minus growth stocks, on
average, are often positive, and significant over long-
term.
•This does not fit well with CAPM.
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Alternative Theories of Asset Pricing

Alternative Theories of Asset Pricing
• CAPM considers investors are rational risk-averse
investors that only consider expected return, risk, and
correlation structure as relevant factors.
• However, investors often behave in irrational manner.
• Arbitrage Pricing Theory (APT) incorporates broad
macroeconomic factors in asset pricing.
• It does not require efficient portfolios.
•𝑅𝑒𝑡𝑢𝑟𝑛 = 𝑎 + 𝑏1 𝑟𝑓𝑎𝑐𝑡𝑜𝑟1 + 𝑏2 𝑟𝑓𝑎𝑐𝑡𝑜𝑟2 + 𝑏3
𝑟𝑓𝑎𝑐𝑡𝑜𝑟3 +⋯+
𝑛𝑜𝑖𝑠𝑒 𝑡𝑒𝑟𝑚
Alternative Theories of Asset Pricing
•The APT theory does not provide any information on
what these factors may be
•One set of risks, that are on account of these APT
factors can not be eliminated with diversification
•PT theory suggests that expected risk premium on a
stock should depend on the
risk-premium associated with each of these
𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑖𝑠𝑘 − 𝑝𝑟𝑒𝑚𝑖𝑢𝑚 = 𝑟
𝑟𝑓𝑎𝑐𝑡𝑜𝑟1 − 𝑟𝑓 +
𝑟𝑓൯𝑟𝑓+ = 𝑏1 (𝑏1, 𝑏2, 𝑏3. . ሻ
• factors and the stock’s
−
𝑟𝑓𝑎𝑐𝑡𝑜𝑟𝑛
sensitivity
𝑏2൫𝑟𝑓𝑎𝑐𝑡𝑜𝑟2 −
⋯ 𝑏𝑛
− 𝑟𝑓
Alternative Theories of Asset Pricing
•As per APT, a well diversified portfolio that is not
sensitive to any risk factor must be priced to offer a
return that is same as risk-free rate.
•A portfolio’s expected return is directly proportional to its
sensitivity to these risk factors.
•A stock’s contribution to a portfolio depends upon its
sensitivity to the broad macroeconomic influences,
often referred to as factors in APT parlance.
•CAPM and APT give similar results if the factors
considered in APT have sensitivity to market portfolio.
Alternative Theories of Asset Pricing
•In CAPM, market portfolio plays a very important
role as it is supposed to capture all the relevant
influences.
•Identifying this portfolio is difficult, however, APT does
not require identification of this market portfolio.
•APT can be tested only with a small number of risky
assets.
•APT does not tell any information about these factors.
•Fama-French three-factor model is a very prominent
example of APT.
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Summary and Concluding Remarks

Summary and Concluding Remarks
• Investors try to increase the expected returns and reduce the
risk on their portfolios.
• A portfolio that gives the highest expected return for a given
standard deviation, or the lowest standard deviation for a
given expected return, is known as an efficient portfolio.
• The best efficient portfolio (tangent) has the highest risk-premium
to standard deviation, i.e., Sharpe ratio.
• As per CAPM, the expected return and risk-premium are
defined by the following model: 𝑅ത𝑖 − 𝑅𝐹 = 𝛽ሺ𝑅ത𝑀 − 𝑅𝐹ሻ
• A stock’s marginal contribution to portfolio risk is measured
by its sensitivity
to changes in the value of the portfolio.
Summary and Concluding Remarks
•The capital asset pricing theory is the best-known
model of risk and return
•However, other risk factors appear to explain the
returns as well
•APT offers an alternative theory of risk and return, i.e.,
expected risk premium depends on the exposure of a
portfolio to various macroeconomic systematic
factors
•One example of APT is Fama-French three factor
model which considers:(a) Market, (b) Size, (c)
Book-to-market (BTM).
Thanks!
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Lesson: Introduction to Portfolio Construction

Advanced Algorithmic Trading and Portfolio
Management
Introduction

• Introduction to portfolio management

• Expected returns and risk for a
portfolio
• Portfolio construction with two-
security case
• Portfolio construction with N-security
case
• Risk diversification with portfolios
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Construction with Two

Securities: Expected Returns
Portfolio Construction with Two Securities

What is a portfolio and why to invest in it?

• What happens to the (1) expected return and (2) risk
when you combine two securities (or multiple
securities)?
• What is diversification?
• Investing in mutual funds and index investing
• What is the difference in risk of investing in Nifty-50
vs. HDFC?
Expected Returns for Two-Security Case

actual return distributions as 𝑅1 and 𝑅2

Consider a portfolio constructed from two-security case with

𝑤1 and 𝑤2, where 𝑤1+𝑤2 = 1

• The proportionate amounts invested in these assets are

• Please also remember that expected returns E(𝑅1) = 𝑅1

and E(𝑅2) = 𝑅2
• Now, let us try to understand the return for the portfolio
• The actual return from the portfolio 𝑅𝑝 (1

• 𝑅 𝑝 = 𝑤1 ∗ 𝑅 1 + 𝑤2 ∗ 𝑅 2
)
Expected Returns for Two-Security Case

What about expected returns?

• E(𝑅𝑝) = 𝐸(𝑤1 ∗ 𝑅1 + 𝑤2 ∗ 𝑅2) (2)
• E(𝑅𝑝) = 𝐸(𝑤1 ∗ 𝑅1) + 𝐸(𝑤2 ∗ 𝑅2) (3)

• E(𝑅𝑝) = 𝑤1 ∗ 𝐸(𝑅1) + 𝑤2 ∗ 𝐸(𝑅2) (4)

= 𝑤1E𝑅1 .
where 𝑤1 and 𝑤2 are constants. Therefore,
E
• However, 𝑅1 and 𝑅2 are probabilistic variables
𝑅1𝑤1finite distributions.
with
Expected Returns for Two-Security Case

What about expected returns?

the probability weightage average. That is, 𝐸(𝑅1)

• For these variables, the expectation operator returns

= 𝑅1; therefore, 𝑅𝑝 = 𝑤1 ∗ 𝑅1 + 𝑤2 ∗ 𝑅2 (5
)
• Expected returns from the portfolio are simply the
weighted average of expected returns of individual
securities in the portfolio.
Expected Returns for Two-Security Case

What about expected returns?

• This can be generalized into three securities and multi-
security as well
𝑅𝑝 = 𝑤1 ∗ 𝑅1 + 𝑤2 ∗ 𝑅2+𝑤3 ∗ 𝑅3,
𝑤1+𝑤2+𝑤3=1
where

• 𝑅𝑝 = σ 𝑁 𝑤𝑖 ∗ 𝑅ഥ𝑖,
“N”σsecurities
𝑁 𝑤𝑖 =
1
• For (6
𝑖=1
where
)
𝑖=1
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Expected Returns from Portfolio: A

Simple Example
Expected Returns: Case 1 (Different Probabilities)

Pt Ra Rb Wa*Ra Wb*Rb 𝑹𝒑=Wa*Ra+Wb*Rb Pt*𝑹𝒑

0.20 9.00% 6.00% 3.60% 3.60% 7.20% 1.44%

0.15 8.00% 5.00% 3.20% 3.00% 6.20% 0.93%

0.10 7.00% 8.00% 2.80% 4.80% 7.60% 0.76%

0.15 11.00% 9.00% 4.40% 5.40% 9.80% 1.47%

0.25 12.00% 10.00% 4.80% 6.00% 10.80% 2.70%

0.15 6.00% 11.00% 2.40% 6.60% 9.00% 1.35%

Wa Wb Total 8.65%

0.40 0.60 E(𝑹𝒑)=P1*𝑹𝒑𝟏+P2*𝑹𝒑𝟐 … … .+P6*𝑹𝒑𝟔

Expected Returns: Case 2 (Equal Probabilities)

Ra Rb Wa*Ra Wb*Rb 𝑹𝒑=Wa*Ra+Wb*Rb

9.00% 6.00% 3.60% 3.60% 7.20%

8.00% 5.00% 3.20% 3.00% 6.20%
7.00% 8.00% 2.80% 4.80% 7.60%
11.00% 9.00% 4.40% 5.40% 9.80%
12.00% 10.00% 4.80% 6.00% 10.80%
6.00% 11.00% 2.40% 6.60% 9.00%
Wa Wb Average 8.43%
0.40 0.60 E(𝑹𝒑)=(1/ N)* (𝑹𝒑𝟏+𝑹𝒑𝟐 … … .+𝑹𝒑𝟔)
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Construction with Two

Securities: Risk
Risk: Standard Deviation for Two Securities

• Variance (σ2)= σ 𝑇 𝑃𝑡(𝑅𝑖,𝑡 − 𝑅ഥ𝑖 )2

The variance of a two-security portfolio
𝑖 𝑡=1

• Again, for past observations that are

• That is, 𝑃1 = 𝑃2 = 𝑃3 = 𝑃4 … … . = 𝑃𝑇 𝑖=
. Since σ 𝑇 𝑃𝑖
equally likely
𝑃 = 𝑃 =
1 we 2have 3𝑃 = 𝑃 4 … =�
1 1

… . = 𝑃𝑇
=1,

)= σ 𝑇 𝑖, − 𝑅𝑖 ഥ
2 1
�
𝑖 𝑡=1
• Variance (σ
𝑡
(𝑅 )2
�
�
Risk: Standard Deviation for Two Securities

The variance of a two-security portfolio

• Think of 𝐴 + 𝐵 2 = 𝐴2 + 𝐵2 + 2𝐴𝐵

• 𝝈𝟐 = 𝒘𝟐 ∗ 𝝈𝟐 +𝒘𝟐 ∗ 𝝈𝟐 + 𝟐 ∗ (𝒘 ∗ 𝝈 )(𝒘
∗𝒑𝝈 )𝝆
𝟏 𝟏 𝟐 𝟐 𝟏 𝟏 𝟐 𝟐 𝟏𝟐
(7)

• where 𝜎𝑝 is the portfolio standard deviation (SD). 𝜎1

and 𝜎2 are SD of the individual securities. 𝑤1 and 𝑤2 are
the investment proportions in each of the securities. 𝜌12 is
the correlation between the two securities, and varies from
−1.0 to 1.0
• What if 𝝆𝟏𝟐=1?
Risk: Standard Deviation for Two Securities

The variance of a two-security

• 𝜎2 = 𝑤 2 ∗ 𝜎2 +𝑤2 ∗ 𝜎2 + 2 ∗ 𝑤 ∗ ∗
portfolio 𝝈𝝆
𝑤∗
∗𝝈
𝑝 1 1 2 2 1 2 𝟏𝟐 𝟏 𝟐
(7
• 𝝆𝟏𝟐 ∗ 𝝈𝟏 ∗ 𝝈𝟐 is called the covariance between securities 1 and 2,
)

also 𝝆𝟏𝟐= 𝝆𝟐𝟏

• This variance (or SD) is less or 2more than the value given by Eq. (8)?
• For 𝝆𝟏𝟐=1, 𝜎2 = 𝑤 ∗ 𝜎+𝑤 ∗ 𝜎
𝑝 1 1 2 2 (8
• 𝝈𝒑 = 𝒘𝟏 ∗ 𝝈𝟏 +𝒘𝟐 ∗ 𝝈𝟐
)

• For all the values of 𝜌12 (except 𝜌12 =1), the value of Eq. (7) will be less
than that of Eq. (8); What are the implications?
Risk: Standard Deviation for Two Securities

• 𝜎2 = 𝑤 2 ∗ 𝜎2 +𝑤2 ∗ 𝜎2 + 2 ∗ 𝑤
The variance of a two-security

∗𝑝 𝑤 ∗ 𝝆 𝟏 ∗ 𝝈
𝟏 �
portfolio
• For 𝝆𝟏𝟐= −1, 𝜎 ∗ 𝜎−𝑤 1 2 2𝟐 ∗ 𝝈
1 2 =2 𝑤 2
(7
1
∗𝜎
� )
𝑝
1
(9
1 𝜌12 (except 𝜌12 = −1), the value of
)
2
• Forall the values of
2
Eq.(7) will be more than Eq. (9); What are the
• 𝝈𝒑 = 𝒘𝟏 ∗ 𝝈𝟏 −𝒘𝟐 ∗ 𝝈𝟐
implications?
Risk: Standard Deviation for Two Securities
𝝈𝟐 = 𝒘𝟐 ∗ 𝝈𝟐 +𝒘𝟐 ∗ 𝝈𝟐 + 𝟐 ∗ ∗ ∗𝝈
𝒘 � 𝟏 𝟏 �
∗𝒑𝒘 �𝝆 𝟐 ∗𝝈
𝟏 𝟏 𝟐 𝟐
1 (𝒘
𝟏
𝟏, 𝝈𝟏) 2 (𝒘𝟐, 𝝈𝟐)
�

1 (𝐰𝟏, 𝛔𝟏) 𝒘𝟐 ∗ 𝝈𝟐 𝝆𝟏𝟐 ∗ 𝒘𝟏*𝝈𝟏*𝒘𝟐*𝝈𝟐

𝟏 𝟏

2 (𝐰𝟐, 𝛔𝟐) 𝝆𝟏𝟐 ∗ 𝒘𝟏*𝝈𝟏*𝒘𝟐*𝝈𝟐 𝒘𝟐 ∗ 𝝈𝟐

𝟐 𝟐
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Construction with Multiple

Securities: Risk
Risk: Standard Deviation for Multiple Securities

𝝈𝟐 = 𝒘𝟐 ∗ 𝝈𝟐 +𝒘𝟐 ∗ 𝝈𝟐 + ∗
𝟐𝒑∗ 𝒘 𝟏 �∗ 𝝆
𝟏 ∗ 𝝈𝟏
𝟏 𝟏 𝟐 𝟐 𝒘 ∗ 𝝈𝟐
𝟐
1 (𝐰𝟏, 𝛔𝟏) 2 (𝐰𝟐, 𝛔𝟐) 3 (𝐰𝟑, 𝛔𝟑)
�

𝝆𝟏𝟐 ∗ 𝝆𝟏𝟑 ∗
1 (𝐰𝟏, 𝛔𝟏) 𝒘 ∗𝝈
𝟐 𝟐
𝟏 𝟏
𝒘𝟏*𝝈𝟏*𝒘𝟐*𝝈𝟐 𝒘𝟏∗𝝈𝟏∗𝒘𝟑∗𝝈𝟑

𝝆𝟏𝟐 ∗ 𝝆𝟐𝟑 ∗
2 (𝐰𝟐, 𝛔𝟐) 𝒘 ∗𝝈 𝟐 𝟐

𝒘𝟏*𝝈𝟏*𝒘𝟐*𝝈𝟐
𝟐 𝟐 𝒘𝟐∗𝝈𝟐∗𝒘𝟑∗𝝈𝟑

3 (𝐰𝟑, 𝛔𝟑) 𝝆𝟏𝟑 ∗ 𝝆𝟐𝟑 ∗

𝒘𝟐 ∗ 𝝈𝟐
𝟑 𝟑
𝒘𝟏*𝝈𝟏*𝒘𝟑*𝝈𝟑 𝒘𝟐*𝝈𝟐*𝒘𝟑*𝝈𝟑
Risk: Standard Deviation for N-Security

1 (𝐰𝟏, 𝛔𝟏) 2 (𝐰𝟐, 𝛔𝟐) ….. N (𝐰𝐍,

𝛔𝐍)
……

1 (𝐰𝟏, 𝛔𝟏)

2 (𝐰𝟐, 𝛔𝟐)

…..

…..

N (𝐰𝐍,
𝛔𝐍)
Risk: Standard Deviation for N-Security

The variance of N-security

𝑖
portfolio
𝑤 𝜎 terms =
• There will be “N” such boxes with entries
2 2 𝑖
2 𝑖=1 𝑤𝑖 𝜎
of
• Variance
𝑖 2
• σ 𝑁
Also, let us assume that all these stocks we have amounts
invested in equal proportion (1/N).
1
𝑁 𝑤𝑖2 𝜎 2 =
• σ 𝑖=1
𝑖=1
𝟏 𝜎𝑖 2 =𝑁1 σ 𝑵
𝒊=𝟏 𝐍 𝒊 because 𝑤
𝑖
=
𝑁
σ 𝑁
𝜎 =
1
𝜎 Variance terms= () ∗2
1 1
𝝈 𝟐 2 𝑁 2 2 𝑖
𝑖=1 𝑁 𝑖 �
, 𝜎avg
avg
• Define
σ𝑁 �
Risk: Standard Deviation for N-Security

The variance of N-security portfolio

• There will also be “𝑁2 − 𝑁” boxes with covariance terms and
cross products of weights invested in both the securities with the
following entries:
𝑖= σ 𝑗= 𝑤𝑖 𝑤
𝑤𝑖 𝑤𝑗 𝜎𝑖 𝜎𝑗 𝜌𝑖𝑗terms = 𝑗 𝑖𝜎𝑗 𝜎𝑖𝑗𝜌 𝑖 , �
𝑤 =𝑤 =
• Covariance
1 𝑁1 also 𝑗 1
σ𝑁 𝑖≠ 1
𝑖= σ 𝑗 (� 2 )𝜎𝑖 𝜎𝑗 =� 2 σ 𝑖= σ 𝑗=1𝜎𝑖 𝜎
�

𝑗
1 � 𝑁1 𝑁𝑖𝑗 𝜌
𝑗=1
• Covariance terms =
σ 𝑁 1 𝑁𝑖≠ �
𝑗
𝜌 𝑖𝑗 𝑖≠𝑗
avg−co = 𝑁(𝑁− σ 𝑖=
1
𝑁 σ𝑗=1
𝑁
𝜎𝑖 𝑗𝜎
• �
v 1 𝑖𝑗
1)
𝜌 𝑖≠𝑗
�
Risk: Standard Deviation for N-Security

The variance of N-security

σ 𝜎𝑖 𝜎
1
𝑖= σ )𝜎 𝜎 𝑁
portfolio
𝑗=1(�
𝑖 𝑗 =� 2 𝑖= 𝑗= 𝑗
1 � σ1𝑁 𝜌𝑖𝑗
• Covariance terms = 2
1 𝑁𝑖≠ �𝜌 1
σ𝑁 𝑗 𝑖𝑗 𝑖≠
avg−co = σ𝑖= 𝑗=1𝜎𝑖 𝜎
1 𝑁
𝑗 𝑗
𝑁(𝑁−
σ1𝑁 𝑖𝑗 𝜌
• �
v 1)
𝑖≠𝑗
�
• σ 𝑖=
𝑁 𝑗=𝜎𝑖 𝜎𝑗 𝜌𝑖𝑗 = Covariance terms*𝑁2 =
1 1
𝜎avg−cov*𝑁(𝑁 − 1)
σ𝑁 𝑖≠
1 2
𝑗 ∗ avg−co =( � ) ∗
�
terms=(𝑁2−𝑁) ∗ 𝜎
• Covariance
v 𝑁−1�
𝜎 avg−cov
�
Risk: Standard Deviation for N-Security

terms=1 ( ) 2∗ 𝜎 ; Covariance
The variance of N-security

)∗𝜎
portfolio
avg−co
• Variance𝑁−1
avg �
v
terms=(
𝑁
1
𝜎2𝑃 = 𝑁
( )
𝑁−1 ∗ 𝜎avg +)(∗
�
•
2
𝑁
𝜎 avg−cov
• Now, if N is very large (N → ∞), then variance term will be
close to zero
• Covariance term will be close to the average covariance

𝜎�2 =
• The portfolio variance will be close to the average
covariance
• 𝜎avg−cov
�

• What are the implications?

 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Risk Diversification with Portfolios

Risk Diversification with Portfolios

• For a well-diversified portfolio

with a large number of
securities, the variance terms
will be close to zero
• Only the average
covariances across the
stocks will contribute to the
portfolio risk
• These covariances arise due
to the correlations between
the security returns
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th edition
• For 7)a portfolio with low correlations across securities, the portfolio
(Chapter
Risk Diversification with Portfolios

• The component associated

with variances is called
diversifiable risk or specific
risk
• Later, we will see that market
does not reward this risk
• The risk that is associated
with covariances is often
called market risk or non-
diversifiable risk
• Market only rewards for
Brealey, Myers and Allen; Principles of Corporate Finance. 10th, 11th, or 12th
bearing
editions. this non-diversifiable
Chapter 7
Example: Computation of Expected Portfolio
Returns
• For example, if we invest 60% of the money in security
1 and 40% of the money in security 2, and the
expected returns from security 1 and security 2 are,
respectively, 8% and 18.8%. Then, the expected returns
from the portfolio are computed as follows:
𝑅𝑝 = 𝑤1 ∗ 𝑅1 + 𝑤2 ∗ 𝑅2
• 𝑅𝑝 = 0.60 ∗ 8.0% + 0.40 ∗ 18.8% = 12.30%
Example: Computation of Expected Portfolio SD

• Consider the same previous example (w1 = 60%, w2 =

compute the portfolio variance: 𝜎1 = 13.2% and 𝜎2 =

40%). Now, some additional information is given to

coefficients: 𝜌12 = −1.0, −0.5, 0, 0.5, and 1. Now, let us

31.0%. Consider five cases of correlation

compute the SD of the portfolio for all the five

• 𝜎2 = 𝑤 2 ∗ 𝜎2 +𝑤2 ∗ 𝜎2 + 2 ∗ 𝑤 ∗ 𝑤 ∗ 𝜌∗ 𝜎
scenarios

∗𝑝 𝜎
1 1 2 2 1 2 12 1 2
Example: Computation of Expected Portfolio SD

Variance (𝛔𝟐)
Case 𝐏 Standard Deviation (𝛔𝐏)
0.62 ∗ 0. 1322 + 0.42 ∗ 0.312 +2 ∗ 0.6 ∗
𝛒𝟏𝟐=1 0.4 ∗ 1
20.32%, which is same
∗ 0.132 ∗ 0.31 = 0.0413
as
= 0.6*13.2%

0.62 ∗ 0. 1322 + 0.42 ∗ 0.312 +2 ∗ 0.6 ∗

+0.4*31.0%
𝛒𝟏𝟐=0.5
0.4 ∗ 0.50
17.74%
∗ 0.132 ∗ 0.31 = 0.0315
0.62 ∗ 0. 1322 + 0.42 ∗ 0.312 +2 ∗ 0.6 ∗
𝛒𝟏𝟐=0.0
0.4 ∗ 0.00
14.71%
∗ 0.132 ∗ 0.31 = 0.0217
0.62 ∗ 0. 1322 + 0.42 ∗ 0.312 +2 ∗ 0.6 ∗
𝛒𝟏𝟐=-0.5
0.4 ∗ −0.5
10.88%
∗ 0.132 ∗ 0.31 = 0.0118
0.62 ∗ 0. 1322 + 0.42 ∗ 0.312 +2 ∗ 0.6 ∗
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Summary and Concluding Remarks

Summary and Concluding Remarks

• Adding more securities that are less correlated

(have lower covariance) in the portfolio leads to
diversification
• Diversification here means the reduction of stock-
specific risk
• The part of the risk that is non-diversifiable is on
account of the covariances across securities
• Often this risk is called market risk or systematic risk
Summary and Concluding Remarks

• Markets do not reward for bearing stock-specific

diversifiable risks
• Since these risks can be easily mitigated, when we say
that we expect certain return for bearing risk, that risk
is systematic/non- diversifiable/market risk
Thanks!
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Lesson: Advanced Portfolio Optimization

Advanced Algorithmic Trading and Portfolio
Management
Introduction

• Portfolio construction: expected returns, risk, correlation, and

covariance
• Portfolio optimization and mean-variance framework: two-
security case and N-security case
• Portfolio possibilities curve and feasible region
• Feasible region with short sales
• Minimum variance portfolio
• Introduction to risk-free lending and borrowing
• Market risk and beta
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Construction Recap I

Expected Returns on a Portfolio

Actual returns on the portfolio can be represented by the

𝑅𝑃𝑡 = 𝑖= 𝑋 𝑅
following model:
𝑖 𝑖
• (1
1
• σ 𝑁
Where 𝑡
‘i’ depicts one of the ‘N’ securities, and ‘Xi’ is the weight )
invested in
the security ‘i’

𝑅 ത = 𝐸 𝑅 𝑖= 𝑋𝑖 𝑅𝑖
• Now, the expected returns of the portfolio can also be written

𝑡)
𝑃 P𝑡
•
1
𝑖= 𝐸(𝑋𝑖 𝑅𝑖𝑡 ) 𝑜𝑟𝑖= 𝑋𝑖𝐸(𝑅𝑖
as:
= ത𝐸( σ 𝑁
𝑁
𝑅 =
•
𝑡)
This can also be written as
σ𝑁 1 1
𝑋𝑖𝑅 𝑖
𝑃
ത σ σ 𝑁
•
follows: (2
𝑖=1 )
Risk of a Two-Security Portfolio

𝜎 =𝐸 𝑅 −𝑅 =𝐸 𝑋 𝑅 ത + 𝑋 𝑅 ത− 𝑋
2ത 𝑅 + 𝑋 𝑅
Risk of a two-security portfolio can be
• 2 2
𝑝 1 1𝑡 2 2𝑡 1 1 2 2
shown as
𝑝𝑡
𝑝
• = 𝐸𝑋1 𝑅1𝑡 + 𝑋2(𝑅2𝑡 − 𝑅ത2) 2
𝑅2𝑡
• = 𝐸−
𝑋1 𝑅𝑅
ത1 − 𝑅ത
1𝑡 1
2 2 + 𝑋2 𝑅
2𝑡 − 𝑅ത2
2 +
− 𝑅ത2 2 ] +
2
• = 𝑋
1 2
𝐸 2𝑋1𝑋2 ]𝑅
2 + 𝑋 2 𝐸[ 𝑅
1𝑡 − 𝑅ത1 2𝑡 − 𝑅ത2
2
2𝑋 𝑋 𝐸[ 𝑅 − 𝑅ത1 𝑅2𝑡
𝑅1𝑡 − 𝑅 1
• The third ത 1
𝑅 2
1𝑡 − 𝑅ത1𝑡
1 𝑅2𝑡 − 𝑅ത2 ]”,
covariance and − 𝑅തbe
term, “𝐸[ is called
2
depicted as 𝜎12 (here 𝜎12 = 𝜎21)
can
Risk of a Two-Security Portfolio

• 𝜎 2 =as
𝑋 2 𝜎 2 + 𝑋 2 𝜎 2 + 2𝑋 𝑋 𝜎
The resulting final expression can be
𝑝 1 1 2 2 1 2 12
shown (3
• This expression can be extended for a three-security )

• 𝜎 2 = 𝑋 2 𝜎 2 + 𝑋 2 𝜎 2 + 𝑋 2 𝜎 2 + 2𝑋 𝑋 𝜎 + 2𝑋 𝑋
portfolio, as shown below

𝜎𝑝 + 2𝑋 𝑋𝜎 (4
1 1 2 2 3 3 1 2 12 1 3 13 1
3 23
)
Few Words on Covariance

Please note that this covariance is the product of two

deviations
𝐸[ 𝑅1𝑡 − 𝑅ത1 𝑅2𝑡 − 𝑅ത2 ]
• If both the securities move together, i.e., positive
deviations and negative deviations are observed for
both securities together, then covariance is expected
to be positive
• Conversely, if positive deviations of one security occur
together with negative deviations of the other security,
then the covariance is expected to be negative
Few Words on Covariance

If the securities do not move together, then the

covariance is expected to be low
• This covariance is standardized in the following manner
to obtain the correlation coefficient, as follows
• 𝝆𝒊 = 𝝈𝒊
𝝈
𝒌𝒊𝝈
(5
𝒌 )
𝒌
• The standardized measure is known as the correlation
coefficient
• It varies between +1 and -1
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Construction Recap II

N-Security Case

Let us start with the variance and covariance expression for a three-
security case.
• 𝝈𝟐 = 𝑿𝟐𝝈𝟐 + 𝑿𝟐𝝈𝟐 + 𝑿𝟐𝝈𝟐 + 𝟐𝑿 𝑿 𝝈 + 𝟐𝑿 𝑿 𝝈
+𝒑𝟐𝑿 𝑿 𝝈
𝟏 𝟏 𝟐 𝟐 𝟑 𝟑
𝟏 𝟐 𝟏𝟐 𝟏 𝟑 𝟏𝟑 𝟏 𝟑 𝟐𝟑

• Terms
• These 𝑋
1 𝜎be
2 2 , called variance
like can segregated 1into two segments
Terms like 2𝑋1𝑋2𝜎12, called covariance terms
terms
• terms
𝑖=1 𝑋𝑖 𝜎
2
• For ‘N’ securities variance, the generalized term can be simply
𝑖
𝑗= σ 𝑘=1(𝑋𝑗 𝑋 𝜎
2
• The as σ 𝑁 [N*(N-1)] term looks like
covariance
written 𝑘
.
1 𝑁𝑗𝑘
this: σ 𝑁
)
𝑗≠𝑘
• 𝛔𝐩𝟐 = σ𝐢=𝟏 𝟐 𝐣=𝟏σ 𝐤=𝟏(𝑿 𝑿
𝒌 𝝈
𝐍
𝐢 𝐗 𝛔
𝟐 𝒋
𝐢 𝑵𝒋≠𝒌 𝒋𝒌
+ σ𝑵 )
N-Security Case: Variance Terms

Assume that we are investing equal amounts in each of

• Then, 𝑋1 = 𝑋2 … . . = �1
these securities

= 𝑋𝑁 1 2
𝑖=1 2 or 1 𝑖= 𝜎
σ
�

� 𝑖
• This means that the variance term will
𝑁
2 𝑁𝜎𝑖 � 1 𝑁
• Assuming the average variance𝑖 of 𝜎ത , the variance
2 σ 𝑁
become
term can
as �1 be
also 𝑖
written
𝜎ത 2�
• For a portfolio with a large number of securities, this variance
term will be closer to zero or very small
N-Security Case: Covariance Terms
1
(Xj Xk σ ( 𝜎𝑗
j=1 σ k=1 𝑗=1 σ 𝑘=1);
What about the covariance
N j≠k
assuming equal investment in each )=σ
jk 𝑁 𝑁 𝑘
σN
𝑁2
term?

• =𝑁−1 𝒋= σ (
security 𝟏
𝑁
𝝈𝒋𝒌)
𝒌=𝟏
𝟏 𝑵 1
σ 𝑵 term 𝑗= σ 𝑘=1 ( 𝜎𝑗 ), is the summation of covariances
𝑵(𝑵−𝟏)
• The
1 𝑁 divided 𝑘by
σ 𝑁
the number of 𝑁(𝑁−1)
covariances: average
covariance (𝜎ത𝑗𝑘)
• Resulting covariance term will become:𝑗
𝑁
𝑁−1
𝜎ത 𝑘

approaches 𝜎ത𝑘𝑗
• As we increase N, this term
N-Security Case

Total standard deviation of

the
𝟏
• 𝝈��
𝒑 = 𝒋
N-security portfolio
𝑵−𝟏�
𝝈 ഥ + 𝝈ഥ 𝒌
𝟐
converges to
•𝑵For 𝒊a large number
�

of securities, this
formula simplifies
•𝝈to𝟐
𝒑
≈𝒋𝒌𝝈ഥ
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition
(Chapter 4)
N-Security Case

• This gives us the intuition

that as the number of
securities is increased,
the variance terms that
represent the risk of
individual securities are
offset
• What is left is that the
covariance terms can
not be diversified away
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition
(Chapter 4)
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Mean Variance Framework

Portfolio Risk and Return Profile

Consider the following equations describing expected

returns and risk from a two-stock portfolio.
• 𝑅𝑝 = 𝑤1 ∗ 𝑅1 + 𝑤2 ∗ 𝑅2 (1
• 𝜎𝑝2 = 𝑤 2 ∗ 𝜎 2 +𝑤 2 ∗ 𝜎 2 + 2 ∗ 𝑤 ∗ 𝜎1 ∗
)
1 1 2 2 1 2 12
∗𝑤
• Consider ∗ 𝜌 𝜎2
two securities 1 and 2. Security 1 offers 8% (2
expected )
return, and 2 offers 18.8% return. SD of 1 is 13.2% and
that of 2
is 31%.
Portfolio Risk and Return Profile

We will examine how the risk-return profile looks for 𝜌12= 1.0
(blue), 𝜌12=0.5 (red), 𝜌12=0 (yellow), 𝜌12=-0.5 (green), and 𝜌12=-
1.0 (black).
Portfolio Risk and Return Profile

We will vary the proportionate amounts, that is, 𝑤1 and 𝑤2,

between 0 and 1 where 𝑤1+𝑤2=1
Portfolio Risk and Return Profile

Consider the blue line with 𝜌12=1 correlation. In this special

case, the equation becomes a straight line: 𝜎𝑝 = 𝑤1 ∗ 𝜎1
+ 𝑤2 ∗ 𝜎2 (blue line)

risk,𝑝𝜎2of
• Across all the graphs, the
lowest
portfolioamount ) for a given
diversification
level
of (highest
return is associated
with the blue line (𝜌12=1)
Portfolio Risk and Return Profile

to 𝜌12 = -1 correlation shown in black

Next, we examine the other extreme case corresponding

• This case (black line)

offers the highest
diversification, as it
carries the lowest levels
of risk for a given level
𝜎 = 𝑤 ∗ 𝜎− 𝑤
of2returns. In this case,
2
∗ 𝜎 equation for risk:
the
𝑝 1 1 2
2
Portfolio Risk and Return Profile

This equation has two solutions, each representing a straight

𝜎𝑝 = (𝑤1 ∗ 𝜎1 − 𝑤2 ∗ 𝜎2 ) when (𝑤1 ∗ 𝜎1 − 𝑤2 ∗ 𝜎2 )>=0;

line: (a)

and 𝜎𝑝 = −(𝑤1 ∗
𝜎1 − 𝑤2 ∗ 𝜎2 ) when (𝑤1 ∗ 𝜎1 − 𝑤2 ∗ 𝜎2 )<0

𝜎𝑝 = 0, 𝑤ℎ𝑒𝑟𝑒 𝑤1 ∗ 𝜎1 = 𝑤2 ∗ 𝜎2 .
• These two lines intersect at

This is a special though

impractical case where we
attained complete
diversification with zero risks
Portfolio Risk and Return Profile

• The cases where 𝜌12 lies between -1 and +1 are

concave kinds of curves in-between the two extreme
cases
• An important observation
here is that the risk of the
portfolio, for a given level of
returns, is sometimes even
less than the least risky
security in the portfolio,
even more so when the
correlation between the
securities is low
Portfolio Risk and Return Profile

Adding more securities to the portfolio surely lowers the

specific risk of the portfolio. Even say 15-20 stocks can
offer a considerable amount of diversification
• What happens when we
add more and more
securities? How does
the feasible region of the
area of possibilities
changes
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Portfolio Possibilities Curve

Portfolio Risk and Return Profile

As we keep on forming these combinations infinitely, we

will get the following convex egg-cut shape.
Portfolio Risk and Return Profile

The region of possibilities is shown in blue

• The blue area is effectively the region of expected return and risk
possibilities that an investor can attain

• Each point represents the

combination of risk and returns
that is available to investors in the
form of investment in portfolios
• Together, all these points
(portfolios) comprise the region of
possibilities (or the feasible region)
How to Improve Our Position in This Region?

We want to move up (increase returns) and move to the left

(reduce risk)
• As we do that, we reach the top surface of the region of
possibilities, that is,
the surface SS’
• There are no more points where we can
move further left or up on this curve (SS’)
• This region would be called the
efficient frontier. And all the points
on this region offer the highest
return for the given level of risk (or
the lowest risk for a given level of
How to Improve Our Position in This Region?

Also, each investor depending upon his risk preference may

choose a specific risk level
• Once he decides on a specific risk level, he will have a given
certain expected return level on the surface SS’
• Once he decides on a specific
risk level, he will have a given
certain expected return level
on the surface SS’
• Two points in this region are
particularly
important for us
How to Improve Our Position in This Region?

Two points in this region are particularly important for us

• Point S has minimum risk as compared to any other point in
the feasible region
• Point S’ that has maximum
return as compared to any
other point on the feasible
region
• All the points between SS’
presents the unique and best
combinations of risk and return on
the feasible region
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Feasible Frontiers
Feasible Frontiers

• The portfolio possibility curve

that lies above the minimum
variance portfolio is concave,
whereas that which lies below
the minimum variance portfolio
is convex
• (b) is not possible because the
combination of assets can not
have more risk than that found
on a straight line connecting two
assets, and that is only the case
where perfect correlation exists
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Feasible Frontiers

• In (c), all the combinations

of U and V must lie on the
line joining U and V or
above such line hence the
given shape is not
possible
• Here, U and V
themselves are
combinations of MV and
C
• Thus, only proper shape
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Feasible Frontiers

• With the same logic as discussed,

MV and any portfolio below MV
(higher variance and lower
return), the resulting curve is
convex
• Thus, both (b) and (c) are not
feasible, only (a) is possible
• Now that we understand the risk-
return properties of combinations
of two assets, we are in a position
to study the attributes of
combinations
Elton, of all risky
Gruber, Brown, and Goetzmann, Modern assets
Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Efficient Frontier Scenarios: Multi-

Security Case I
 Efficient Frontier Scenarios: Multi-Security
Case

• Efficient frontier with no-short sales

• Efficient frontier with short sales (no risk-free
lending and borrowing)
Efficient Frontier with No-Short Sales: Multi-
Security Case
In this diagram, we try to find portfolios
that offer a higher returns for a given
level of risk or
• Offered a lower risk for the same
return
• Here, portfolio B would be
preferred over portfolio A, and
portfolio C would be preferred over
Assume C to be minimum
portfolio A variance (MV)
portfolio
• No portfolio dominates a portfolio And B to be maximum return
such as B or C portfolio
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Efficient Frontier with No-Short Sales: Multi-
Security Case
• C here is the global
minimum variance
portfolio
• Portfolio E is superior to
portfolio F
• Thus, efficient sets of
portfolios are those that lie Assume C to be minimum variance
between the global minimum (MV) portfolio
And B to be maximum return
variance portfolio and the portfolio

maximum return portfolio

Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Efficient Frontier with No-Short Sales: Multi-
Security Case
• The efficient frontier here A
is a concave curve (A)
• Why should it be a concave
curve (not convex like the
segment between U and V
on B)? B

• In this case (A), the efficient

frontier (EF) is a concave
function; EF extends from
minimum variance portfolio to
maximum return portfolio
Efficient Frontier with Short Sales
• With short sales, one can sell
securities with low expected
returns and use the proceeds to
buy securities with high expected
returns
• Theoretically, this leads to infinite
expected rates of return but
extremely high standard
deviations as well
• MVBC becomes the efficient frontier
which is concave
• The efficient set still starts with the minimum variance
portfolio, but when short sales are allowed, it has no
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Efficient Frontier Scenarios: Multi-

Security Case: II
Efficient Frontier Scenarios: Multi-Security
Case
• Efficient frontier with riskless lending and
borrowing
• Only riskless lending is allowed; not
borrowing
• Riskless lending and borrowing at different
rates
Efficient Frontier with Riskless Lending and
Borrowing
• Introduction of riskless assets
considerably simplifies the
analysis
• Tangent line from 𝑅𝐹 to G offers
a new set of the efficient
portfolios with a maximum
(𝑅𝐺σ−𝑅𝐹)
expected return premium for a
𝐺 ; where G is the
given level of risk
portfoli
tangent
o
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Efficient Frontier with Riskless Lending and
Borrowing
• Very risk-averse investors would

investment in risk-free assets: 𝑅𝐹

hold portfolio G along with some

−G (lending portion)

would borrow some amount at 𝑅𝐹

• Those who are more risk-tolerant

and invest the entire money in the

𝐺 − 𝐻 (borrowing portion)
tangent portfolio (G):

• Separation theorem:
identification of optimum
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
portfolio
Chapter 5 does not require
Only Riskless Lending Is Allowed;
Not Borrowing
• If investors can lend but not
borrow at the risk-free rate,
then the efficient frontier

𝑅𝐹 − 𝐺 − 𝐻
becomes

• Some investors will hold 𝑅𝐹

𝑅𝐹 − 𝐺), and others will hold

and G (positioned on the line

a risky portfolio between G

and
Elton, Gruber,HBrown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
Riskless Lending and Borrowing at Different
Rates
• Another possibility is that
investors can lend at one
rate but must pay a
different and presumably
higher (𝑅𝐹 to �
borrow rate
• and 𝑅 ′ )
The efficient �

𝑅𝐹 − 𝐺 − 𝐻
frontier

−𝐼
Elton, Gruber, Brown, and Goetzmann, Modern Portfolio Theory and Investment Analysis, 9th edition,
Chapter 5
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Minimum Variance Portfolio

Minimum Variance Portfolio

In the absence of short sales, two points become extremely

important on the efficient frontier
• First, the portfolio with maximum return, and second the
minimum variance portfolio
• In the absence of short sales, these portfolios define the two
extreme ends of the efficient frontier
• While it is easy to understand that a maximum return portfolio will
be the
security in the portfolio that offers the maximum return
• The same is not the case for minimum variance portfolio
Minimum Variance Portfolio

This portfolio is often expected to be different from the security

with minimum risk (SD) in the portfolio. How do we compute this
1
• 𝜎𝑃 = 𝐴 𝑋 𝐴 𝜎 𝐵+ 𝑋 𝜎 + 2𝑋𝐴𝑋𝐵𝜌 𝐴𝐵𝜎𝐴𝜎𝐵
portfolio?
2 2 2 2 2 (1)
𝐵
• What exactly do we want to compute here?
• 𝜎𝑃 = 𝑋2 𝜎2 + 1 − 𝑋𝐴 2 𝜎2 + 2𝑋 1 − 𝑋𝐴 (2
1
𝐴 2
𝐴 𝐴 𝐵 𝜌𝐴𝐵𝜎𝐴

𝜎𝐵
)
• To obtain the minima, we need to set the derivative = 0 in Eq. (2),

𝑋
and solving
• 𝑋� =
this for 𝜎 �2 𝐴, we get
𝜎𝐴−𝜌
+𝜎𝐴 𝐵 𝜎𝐴 𝜎
−2𝜌 𝐴 𝐵𝜎 𝐴
2 � 2
(3
𝐵
𝜎𝐵
� )
Minimum Variance Portfolio

Consider the example

below Stock Expected Returns SD
A 14% 6%
B 8% 3%

• Assume a correlation of 0, try to find the amount invested in

= 2 2 = 2 2 = = 0.2 , 𝑋� =
𝜎�2
MV portfolio
𝜎𝐵2 3 2 1
�
• 𝑋�= 𝜎𝐴 0.8
2−𝜌
+𝜎𝐵𝐴 𝐵
� 2
𝜎𝐴 𝜎
−2𝜌 𝐴𝐵𝐵𝜎 𝐴 𝜎𝐴+𝜎 6 +3
𝐵 15
𝜎𝐵
• 𝜎𝑃 0.22 ∗ 62 + 0.8232 2 =
�

= 2.68%
Minimum Variance Portfolio

Consider the example

below Stock Expected Returns SD
A 14% 6%
B 8% 3%

• Assume a correlation of 0.5, try to find the amount invested

• 𝑋� = = 2 2
𝜎�2 −𝜌 𝐴 𝐵 𝜎 𝐴 𝜎 𝐵
=
0
𝜎𝐴23+𝜎 2 −2𝜌
2−0.5∗6∗3
�
𝐴 𝐵 𝜎𝐴
6 +3 −2∗0.5∗6
∗3
• What 𝜎is𝐵𝐵 the implication?
�
No combination of securities A and B
has less risk than security B itself. So, the minimum variance
portfolio is security B itself. That also means for any correlation

portfolio (𝑋𝐴 = 0)
higher than 0.5, security B will itself B the minimum variance
Minimum Variance Portfolio

Consider the example

below Stock Expected Returns SD
A 14% 6%
B 8% 3%

• Assume a correlation of 1, try to find the amount invested

• 𝑋� = 𝜎�2
= <
𝜎 𝐵 (𝜎 𝐵 −𝜎 𝐴 )
=
0
𝜎𝐴−𝜌
2 +𝜎𝐴 𝐵
� 2
𝜎𝐴 𝜎
−2𝜌 𝐴𝐵𝐵𝜎 𝐴 𝜎 𝐵 𝜎 𝐴 −𝜎 𝐵 𝜎 𝐵 −𝜎
2𝐵 𝜎𝐵
�
𝐴
• With limiting constraint that any weight cannot be equal to zero,

𝑋𝐴 = 0
this gives us
Minimum Variance Portfolio

Consider the example

below Stock Expected Returns SD
A 14% 6%
B 8% 3%

• Assume a correlation of -1, try to find the amount

𝜎�2 𝜎 𝐵 (𝜎 𝐵 +𝜎 𝐴 ) 𝜎𝐵 = 1/3
= = � =
invested
�
• 𝑋�= 1 𝐵
𝜎𝐴−𝜌+𝜎𝐴 𝐵 𝜎𝐴 𝜎
−2𝜌 𝐵𝜎 𝐴 𝜎 𝐴 +𝜎 𝐵
,𝑋
𝜎𝐵 +𝜎𝐴
2 � 2 2
𝐴𝐵
2
• 𝜎 = 𝜎∗ 𝐵 6 −
2
∗3 3
�

=�0
�
3
3
• 𝑊𝐴 𝜎𝐴 − WB𝜎𝐵=0
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Introduction to Risk-Free Lending and

Borrowing I
Introduction to Risk-Free Lending and
Borrowing

of interest 𝑟𝑓
Let us introduce risk-free lending and borrowing at the risk-free rate

• What are the practical challenges with this assumption

• Can we borrow at the same rate from the State Bank of India (SBI)
at which
we make fixed deposits with SBI
• However, this assumption has several important implications
for portfolio construction
• Consider that a large number of stocks are employed to
construct a feasible region of possibilities
Introduction to Risk-Free Lending and
Borrowing
In practice, you invest in a portfolio of number
of stocks
• Thus, you obtain a wider
selection of risks and return
• You also obtain the efficient
frontier by going up (increase
expected return) and to the
left (reduce risk)
• This becomes a capital
rationing problem, which can
be solved with quadratic
programming
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
Chapter 8
The Efficient Frontier with Riskless Lending
and Borrowing
• The addition of
riskless securities
considerably
simplifies the analysis
and opens new
possibilities for
investment
• Consider two investments
(1) a portfolio of assets A
that lies on the efficient
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
frontier;
Chapter 8 and (2) one risk-
The Efficient Frontier with Riskless Lending
and Borrowing
If X fraction of the amount is
placed in the portfolio, then 1 −
X fraction will be placed in the
riskless asset
• The expected return on this
portfolio can be expressed
by the following equation:
(1
• 𝑅
𝜎
ത𝑝==𝑋𝑋2𝑅
𝜎 +
ത2𝐴 + 1 − 𝑋
1 −𝑅𝑋 2 𝜎2 + 𝐴𝑓𝜎𝐴
)
𝑓
•
2 � 2𝑋 𝑓 1−𝑋 𝜌 𝑓 𝜎
𝐴
�
(2
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
)
Chapter 8
The Efficient Frontier with Riskless Lending
and Borrowing
The equation for risk can be simplified with the introduction
of risk-free instrument
• 𝜎 = 𝑋2 𝜎2 + 1 − 𝑋 2 𝜎 2 + 2𝑋
𝑓
1 − 𝑋𝜎 =𝜌 0, the 𝜎 following
𝜎
𝑝 𝐴 𝐴𝑓 𝐴
2
𝑓
• Because 𝑓 expression of the portfolio risk
is obtained
• 𝜎𝑝 = 𝑋𝜎𝐴
(1
)

• 𝑅ത ത𝑝𝑝 == 𝑅
𝑋𝑓𝑅+ 𝐴 (+ 1)𝜎
ത𝑅ത 𝐴−𝑅 − 𝑋 𝑅𝑓
(2
𝜎
•
)
ത𝑓
𝐴 𝑝
• This (Eq. 3) is the equation of a straight line that passes (3
through all the combinations of riskless lending or borrowing )
with portfolio A
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Introduction to Risk-Free Lending and

Borrowing II
Introduction to Risk-Free Lending and
Borrowing
The brown line represents the
most efficient portfolios or the
efficient frontier
• Now that you have risk-free
asset, you can invest a certain

investment at 𝑟𝑓 and the

amount in the risk-free

remaining amount on any

portfolio available on the
surface “S” corresponding to
the efficient frontier
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
Chapter 8
Introduction to Risk-Free Lending and
Borrowing

from the point 𝑟𝑓 to the red

Let us draw a line tangent

line curve
• The line that is the steepest
among all is the tangent line
• The slope of this line is the
amount
of return per unit of risk. That
𝑟 𝑆 −𝑟 𝑓
is,

𝜎𝑝

• This
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
Chapter 8
Introduction to Risk-Free Lending and
Borrowing
Now, we have an even better
position, which is shown by the
line going through rf and rs
• It has two segments
borrowing and lending for
investors with high and low-
risk preference

𝑟𝑓 and investing at 𝑟𝑆 is
• This strategy of borrowing at

depicted by the line segment

called borrowing
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
Chapter 8
Introduction to Risk-Free Lending and
Borrowing

𝑟𝑓 and partially at 𝑟𝑆 ,
I can invest partially at

and hold a portfolio on

the line segment called
lending
• If the portfolio S is
known with reasonable
certainty, everybody
should hold this
portfolio, and this will
Brealey, Myers, and Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
be called market
Chapter 8
Introduction to Risk-Free Lending and
Borrowing
In a competitive market,
everybody is expected to hold
this market portfolio, and the job
of the investment manager is
expected to be fairly easy
• One must identify the
market portfolio of
common stocks
• Then mix this portfolio with
risk- free lending or borrowing
to create a product that suits
theMyers,
Brealey, tasteand and risk preference
Allen, Principles of Corporate Finance, 10th, 11th, or 12th editions,
Chapter 8
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Introduction to Risk-Free Lending and

Borrowing III
Introduction to Risk-Free Lending and
Borrowing: Simple Example
Suppose market portfolio S here offers 15% expected returns and
SD of 16%. The risk-free instrument offers a 5% uniform rate of
lending and borrowing, with an SD=0.
You are a risk-averse investor; therefore, you would like to invest
50% into rf and balance into S. What does your portfolio look like.
The corresponding equations for the risk and expected returns on
the portfolio are provided below
𝜎𝑝 = 𝑋𝜎𝐴

𝑅ത𝑝 = 𝑋𝑅ത𝐴 + 1 − 𝑋 𝑅𝑓
Introduction to Risk-Free Lending and
Borrowing: Simple Example
You are a risk-averse investor; therefore, you would like to invest
50% into rf and balance into S.
𝝈𝒑 = 𝑿𝝈𝑨

𝑅ത𝑝 = 𝑋𝑅ത𝐴 + 1 − 𝑋𝑅𝑓

𝑟𝑓 ∗ 0.5 + 𝑟𝑆 ∗ 0.5 = 5% ∗ .5 + 15% ∗ 0.5 = 10%.

The expected returns on your portfolio are

The standard deviation of the portfolio will be 𝜎𝑝 = 0.5 ∗ 16% =

8%.
You are standing on the lending segment of the line of
investment at a point, that is, midway between rf and rs.
Introduction to Risk-Free Lending and
Borrowing: Simple Example
• Another investor who is more risk-taking in his approach
will borrow at rf almost 100% and invest 200% in the

shown below. His return will be 𝑟𝑓 ∗ −1.0 + 𝑟𝑆 ∗ 2.0 =

market portfolio. The risk-return profile of this investor is

5% ∗ −1.0 + 15% ∗ 2.0 = 25%. At the same time, his

risk will be 𝜎𝑝 = 2 ∗ 16% = 32%.
• This investor has extended his possibilities and
operates on the borrowing segment of the line.
Introduction to Risk-Free Lending and
Borrowing: Simple Example
So, whether it is fearful chickens or risky lions, both will prefer
this market portfolio as compared to any of the portfolios on
the efficient frontier
• Therefore, this market portfolio is the best efficient portfolio for
the entire set of investors
• And we also know how to identify this portfolio by drawing a
tangent line from rf to on the surface of efficient portfolios
discussed 𝑅𝑖𝑠𝑘 𝑟𝑆−𝑟highest risk
𝑃𝑟𝑒𝑚𝑖𝑢𝑚
𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝜎
• standard
This portfolio, as weSharpe
deviation: earlier, offers the
=
𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
premium
ratio: to the
𝑓 𝑝
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Market Risk and Beta

Market Risk and Beta

Market risk is the risk associated with a well-diversified

portfolio, often called a market portfolio (Nifty 50)
• If a sufficiently large number of securities are added to a
portfolio, the only risk that remains is the
non-diversifiable/systematic/market risk
• What is this market risk?
• The contribution of a security to the portfolio is
determined by the correlation of a security (or the
covariance) with the market portfolio
Market Risk and Beta

market portfolio is represented through beta (𝛽𝑖)

This correlation or the sensitivity of the security (i) with the

• For example, if security moves by 1.5% for a 1% movement in

the market portfolio, then the beta of a security is said to be
1.5
• If the beta of a security is 1.0, then security is said to be having
same risk as that of the market
• If beta is 0, then the security doesn’t have any market risk:
government
securities
• In summary, this beta represents the sensitivity of the
security to market movements
Market Risk and Beta

Beta of a portfolio is weightage average betas of the individual

securities

(𝛽1, 𝛽2, 𝛽3, 𝛽4 … . . 𝛽𝑁) and proportionate amounts invested in

• For example, if we have N securities with individual betas

securities are 𝑤1, 𝑤2 … . . 𝑤𝑁. 𝑖=

𝛽𝑃 = 𝑤1 ∗ 𝛽1 + 𝑤2 ∗ 𝛽2 … 𝑤𝑁 ∗ 𝛽𝑁 = σ 𝑤𝑖 ∗ 𝛽𝑖
• these 𝑁
Then, the beta of the
portfolio can be written as below 1
• If the observed standard deviation of the market is 20%. Now we
construct a portfolio from a large number of securities with an
average beta of 1.5
• The standard deviation of this portfolio will be 30% (1.5*20%)
Market Risk and Beta

Beta of individual security (𝛽𝑖) is defined and computed

as𝛽follows.
𝑖 = 𝜎 𝑖𝑚𝑚/𝜎 2 ; here, 𝜎
𝑖𝑚 is the covariance between
the market returns (expected). 𝜎𝑚 is the standard
•
and security
the
deviation of the expected market returns
• How to compute betas in real life
• Returns of the security are regressed on the market
returns. Market returns can be proxied using broad
indices such as Nifty, NYSE
Market Risk and Beta: Regression Analysis
Example: Beta Computation
A B C D E F

𝝈𝟐 = (𝑹 − 𝑹ഥ )𝟐 𝝈𝒊𝒎 =
𝑹𝒎 𝑹𝟏 𝑹𝒎 − 𝑹ഥ𝒎 𝑹𝟏 − 𝑹ഥ𝟏
𝑹𝟏 − 𝑹ഥ𝟏 ∗ (𝑹𝒎 − 𝑹ഥ𝒎)
Period 𝒎 𝒎 𝒎

1 -1.00 3.60 -1.30 -0.10 1.69 0.14

2 -6.00 3.20 -6.30 -0.50 39.69 3.18
3 10.00 4.48 9.70 0.78 94.09 7.53
4 10.00 4.48 9.70 0.78 94.09 7.53
5 -3.00 3.44 -3.30 -0.26 10.89 0.87
6 -11.00 2.80 -11.30 -0.90 127.69 10.22
7 8.00 4.32 7.70 0.62 59.29 4.74
8 -6.00 3.20 -6.30 -0.50 39.69 3.18
9 10.00 4.48 9.70 0.78 94.09 7.53
10 -8.00 3.04 -8.30 -0.66 68.89 5.51

𝜎𝑖𝑚 = 5.04, 𝜎 2 = 63.01,

Avg. 0.30 3.70 63.01 5.04

𝑚 =
𝑖
𝑎𝑛𝑑 𝛽 =
�
𝑚 𝜎𝑚 2
0.08
�
𝑖
 INDIAN INSTITUTE OF TECHNOLOGY
KANPUR

Summary and Concluding Remarks

Summary and Concluding Remarks

• For a portfolio with a large number of securities, only systematic

(or market) risk is relevant
• Idiosyncratic stock-specific risk is eliminated due to
diversification
• When two securities are perfectly correlated 𝜌12 = 1, no
diversification is achieved
• When two securities are perfectly negatively correlated 𝜌12 = -
1, maximum diversification is achieved
• As we keep on adding more and more securities, the region of
all possible risk-return scenarios is obtained (feasible region)
Summary and Concluding Remarks

• On this feasible region, we would like to go up (increase

expected returns) and go to the left (decrease the
risk)
• When short-selling is not allowed, a set of best efficient
portfolios from minimum variance portfolio to maximum
return portfolio are obtained that dominate all other
risk-return profiles: efficient frontier (EF)
• When short-selling is allowed, an extended feasible
region is obtained, the efficient frontier is also
extended on the top-right
Summary and Concluding Remarks

• In the presence of risk-free security, a new efficient

frontier is obtained, which is a tangent line joining risk-
free security to the tangency point
• On this new efficient frontier, the line segment toward
the left of the tangency point is called the lending
segment: a mix of investment into risk-free security
and tangency portfolio
• The line segment towards the right of the tangency
point is called the borrowing segment: borrowing at the
risk-free rate and investing the complete amount into
the tangency portfolio
Thanks!