QF2104 Fundamentals of

Quantitative Finance

Dr Chunchun Liu

1
Chapter 4: Mean-variance Analysis

Return and risk of asset

➢ Single asset
➢ Portfolio
Portfolio of 2 assets
➢ Feasible sets
➢ Global minimum-variance portfolio
Portfolio of 3 or more assets
➢ Minimum-variance frontier
➢ Efficient frontier
2
Return and Risk of Asset

The main purpose of this chapter is to discuss the concept of

investment risk and return which lay the foundation for
Portfolio Theory (to be discussed in next Chapter).

➢ We assume that there are finitely many tradable (can be

readily bought and sold) financial instruments called
assets in the financial market

➢ We assume that each asset is traded over one time

period, from t = 0 (initial) to t = 1 (end-of-period).

3
Rate of Return of Asset

If $𝑊0 invested in an asset at 𝑡 = 0 is worth a random

amount of $𝑊1 at time 𝑡 = 1, then the rate of return of the
asset, denoted by 𝑟, is a random variable given by
𝑊1 − 𝑊0 𝑊1
𝑟= = −1
𝑊0 𝑊0

Equivalently, 𝑊1 = 𝑊0 (1 + 𝑟)

4
Rate of Return of Asset

The rate of return can also be defined in terms of the initial

and end-of period prices of the asset. Let 𝑃0 be the price at
𝑡 = 0 and 𝑃1 be the random price at 𝑡 = 1. Then
𝑃1 − 𝑃0 𝑃1
𝑟= = −1
𝑃0 𝑃0

Equivalently, 𝑃1 = 𝑃0 (1 + 𝑟)

5
Return and Risk of Asset

The mean value 𝜇 = 𝐸(𝑟𝑖 ), of the rate of return of asset 𝑖, is

a measure of the return of asset 𝑖.

The standard deviation 𝜎𝑖 = 𝑉𝑎𝑟(𝑟𝑖 ), of the rate of return of

asset 𝑖, is a measure of the risk of asset 𝑖.

6
Correlation of Returns

A statistical measure of the association of the returns of the

returns of two assets, 𝑖 and 𝑗, is the covariance 𝜎𝑖𝑗 =
𝑐𝑜𝑣(𝑟𝑖 , 𝑟𝑗 ).

A standardized measure is the correlation coefficient defined

by

𝜎𝑖𝑗
𝜌𝑖𝑗 ≡
𝜎𝑖 𝜎𝑗

It can be shown that 𝜌𝑖𝑗 ≤ 1.

7
An Example

A hypothetical end-of-period prices of shares of company ABC are

given below:
Share price
3.50 4.20 5.00 5.50 6.00

Probability
0.15 0.10 0.30 0.20 0.25

The initial share price is $5.00.

(i) A man buys 200 shares of ABC. Calculate the expected profit/loss,
mean rate of return and standard deviation of the rate of return.

(ii) Suppose the man buys 100 shares of ABC and $x worth of
another security whose mean rate of return is 2%. If his one period
mean rate of return is 1.725%, find x.

8
Solution

9
Short Selling

When an individual short sells (or simply shorts) an asset, he

borrows a certain number of units of the asset from another
individual (the lender) at 𝑡 = 0 and sells them immediately to
receive an amount $𝑊0 . At some pre-agreed date 𝑡 = 1, the short-
seller will buy the same number of unit of the asset for an amount
$𝑊1 and return the asset to the lender. Assuming there are no
transaction costs, the borrower makes a profit of $(𝑊0 − 𝑊1 ) which
is positive iff the value of the asset falls.
Note that the loss from short-selling an asset can be potentially
unlimited (theoretically, the asset value 𝑊1 can be infinitely large)
while the gain is bounded above by 𝑊0

10
Short Selling

𝑆ℎ𝑜𝑟𝑡 𝑠𝑎𝑙𝑒 𝑔𝑎𝑖𝑛

= #𝑠ℎ𝑎𝑟𝑒𝑠 𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑡 = 0 − (𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑡 = 1)
= − #𝑠ℎ𝑎𝑟𝑒𝑠 𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑡 = 1 − (𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑡 = 0)

11
Margin Deposit

The broker requires that the investor provide some collateral in the
form of a margin deposit to protect against the risk of default by the
investor. The broker is at risk if the stock price increases and the
investor is unable to afford replacing the stock. Usually the margin
deposit is expresses as a percentage of the initial share price.

12
Dividends

Along with the obligation to replace the borrowed shares, the

short seller must also pay the broker for any dividends paid on
the stock during the period of the short sale

Short sale profit = short sale gain + margin interest – dividends

𝑠ℎ𝑜𝑟𝑡 𝑠𝑎𝑙𝑒 𝑝𝑟𝑜𝑓𝑖𝑡

Short sale return rate/yield =
𝑚𝑎𝑟𝑔𝑖𝑛 𝑑𝑒𝑝𝑜𝑠𝑖𝑡

13
An Example

The initial stock price per share is $50, and the price is $45 one
year later. Suppose that the investor short sells 40 share, the
broker requires 50% as the margin deposit, the effective annual
interest rate is 4%, and each share of stock is paid a dividend of
0.5 at the end of the year.

(i) Find the gain on the short sale.

(ii) Find the margin deposit at 𝑡 = 0.

(iii) Find the short sale profit.

(iv) Find the short sale return rate.

14
Solution

15
An Example
An investor decides to short sell 500 shares of ABC stock. The current price per
share is $48.25. The broker requires 50% margin deposit. The effective annual
interest rate is 2.5%.
(i) If the stock declines by 20% over the next year and he closes out the short
position in exactly one year, what will the rate of the return be if no
dividends are declared.
(ii) If the stock declines by 20% over the next year and he closes out the short
position in exactly one year, what will the rate of the return be if a dividends
of $0.15 per share is declared right before the short position is closed.
(iii) If the stock rises by 20% over the next year and he closes out the short
position in exactly one year, what will the rate of the return be if no
dividends are declared.
(iv) If the stock rises by 20% over the next year and he closes out the short
position in exactly one year, what will the rate of the return be if a dividends
of $0.15 per share is declared right before the short position is closed.
16
Solution

17
An Example

An investor initiates a short sale for 400 shares when the per share
price is $43.13. she closes out her short sale position exactly one
year later, buying back the stock for $38.95 per share. The initial
margin requirement is 55%. Assume that there are no intermediate
margin deposit except for quarterly withdrawals of $0.36 per share
to cover dividends. The annual effective interest rate is 2.82%. Find
the investor’s rate of return for this one-year transaction.

18
Solution

19
Simplification for Short Selling

In the rest of Chapter 4 and Chapter 5, we assume for

simplicity that there is no margin deposit and no dividends

for short sales.

➢ Profit = short sale gain

➢ Rate of return is not well defined

20
Portfolio Mean and Variance

At time t = 0, an individual invests in n assets in such a way that a

fraction 𝑤𝑖 of his investment capital is invested in asset 𝑖. It is possible
that 𝑤𝑖 < 0, which means the individual short sells asset 𝑖. We call the
vector 𝒘 = (𝑤1 , 𝑤2 , … , 𝑤𝑛 )𝑇 the individual’s portfolio weight vector
(or simply, portfolio). We shall assume, unless stated otherwise, that
𝑛

෍ 𝑤𝑖 = 1
𝑖=1

The rate of return, 𝑟𝑝 of the portfolio is related to the rate of return of

individual asset , 𝑟𝑖 by
𝑛

𝑟𝑝 = ෍ 𝑤𝑖 𝑟𝑖
𝑖=1
21
Portfolio Mean and Variance

It follows that the mean rate of return of the portfolio, or simply,

portfolio mean, is
𝑛

𝜇𝑝 = 𝐸 𝑟𝑝 = ෍ 𝑤𝑖 𝜇𝑖
𝑖=1
In matrix notation, we have
𝜇𝑝 = 𝒘𝑻 𝝁
where
𝝁 = (𝜇1 , 𝜇2 , … , 𝜇𝑛 )𝑇
is the vector of mean rates of return of the assets. We shall call this
vector the mean vector for simplicity.
22
Portfolio Mean and Variance

As for the variance of the rate of return, or simply portfolio

variance, of the portfolio 𝒘 = (𝑤1 , 𝑤2 , … , 𝑤𝑛 ), we have

𝑛 𝑛

𝜎𝑝2 = 𝑉𝑎𝑟 𝑟𝑝 = ෍ ෍ 𝑤𝑖 𝑤𝑗 𝜎𝑖𝑗

𝑖=1 𝑗=1

Since 𝑉𝑎𝑟 𝑟𝑝 = Var σ𝑛𝑖=1 𝑤𝑖 𝑟𝑖 = Cov σ𝑛𝑖=1 𝑤𝑖 𝑟𝑖 , σ𝑛𝑗=1 𝑤𝑗 𝑟𝑗 =

σ𝑛𝑖=1 σ𝑛𝑗=1 𝑤𝑖 𝑤𝑗 𝜎𝑖𝑗

23
Portfolio Mean and Variance

In matrix notation, we have

𝑉𝑎𝑟 𝑟𝑝 = 𝑾𝑻 𝑪𝑾
where

𝜎1,1 , 𝜎1,2 , … , 𝜎1,𝑛

𝜎2,1 , 𝜎2,2 , … , 𝜎2,𝑛
𝐶= ⋮
𝜎𝑛,1 , 𝜎𝑛,2 , … , 𝜎𝑛,𝑛

Is known as the covariance matrix of the random vector 𝑟 =

𝑇
𝑟1 , 𝑟2 , … , 𝑟𝑛

24
Portfolio Mean and Variance

Since 𝜎𝑖𝑖 = 𝜎𝑖2 = 𝑉𝑎𝑟(𝑟𝑖 ) and 𝜎𝑖𝑗 = 𝜎𝑗𝑖 , we have

𝑛 𝑛

𝑉𝑎𝑟 𝑟𝑝 = ෍ 𝑤𝑖2 𝜎𝑖2 + ෍ ෍ 𝑤𝑖 𝑤𝑗 𝜎𝑖,𝑗

𝑖=1 𝑖=1 𝑗≠𝑖

𝑛 𝑛

= ෍ 𝑤𝑖2 𝜎𝑖2 + 2 ෍ ෍ 𝑤𝑖 𝑤𝑗 𝜎𝑖,𝑗

𝑖=1 𝑖=1 𝑗<𝑖

25
An Example

A financial market consist of 3 risky assets whose rates of return 𝑟1 , 𝑟2 and

1 1 𝑇
𝑟3 have mean vector 𝝁 = , ,1 and covariance matrix 𝑪 defined by
2 2

2 𝑖𝑓 𝑖 = 𝑗
𝜎𝑖𝑗 = ቐ−1 𝑖𝑓 𝑖 + 𝑗 = 5
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Portfolio 𝑝 is the equally weighted portfolio whose weight vector is 𝑤𝑝 =
1 1 1 𝑇
, , . Portfolio 𝑞 is equally weighted in assets 1 and 2, i.e. 𝑤𝑞 =
3 3 3
𝛼, 𝛼, 1 − 2𝛼 𝑇 .
(i) Find the mean and variance of the portfolio 𝑝.
(ii) If short-selling is not allowed, find the largest possible value of 𝜇𝑞 .
26
Solution

27
Portfolio of Two Assets

𝑡
Consider a portfolio 𝑤 = 𝛼, 1 − 𝛼 of two assets. The portfolio mean and
variance of 𝑤 are given respectively by
𝜇𝑝 = 𝛼𝜇1 + (1 − 𝛼)𝜇2

and
𝜎𝑝2 = 𝛼 2 𝜎12 + (1 − 𝛼)2 𝜎22 + 2𝛼 1 − 𝛼 𝜎12
= 𝛼 2 𝜎12 + (1 − 𝛼)2 𝜎22 + 2𝛼 1 − 𝛼 𝜌12 𝜎1 𝜎2

A risk averse individual desires a portfolio with the smallest risk. He will
thus seek the optimal value 𝛼 that minimizes 𝜎𝑝2 .

28
Global Minimum-variance Portfolio

It can be shown that the minimum portfolio variance 𝜎𝑝2 occurs when

𝜎2 (𝜎2 − 𝜌1,2 𝜎1 )
𝛼 = 𝛼∗ =
𝜎12 + 𝜎22 − 2𝜌1,2 𝜎1 𝜎2

and the minimum portfolio variance is

∗ 𝜎12 𝜎22 (1 − 𝜌1,2

2
)
𝜎𝑝2 = 2
𝜎1 + 𝜎22 − 2𝜌1,2 𝜎1 𝜎2

The corresponding portfolio mean can be determined from

∗
𝜇𝑝 = 𝛼 ∗ 𝜇1 + 1 − 𝛼 ∗ 𝜇2

We call the portfolio with minimum variance the global minimum-

variance portfolio (GMVP).
29
Portfolio Graph (Graph of 𝝁𝒑 vs 𝝈𝒑 )

Combining
𝜇𝑝 = 𝛼𝜇1 + (1 − 𝛼)𝜇2

and
𝜎𝑝2 = 𝛼 2 𝜎12 + (1 − 𝛼)2 𝜎22 + 2𝛼 1 − 𝛼 𝜌12 𝜎1 𝜎2

we can obtain an equation of the form

𝜎𝑝2 = 𝐴𝜇𝑝2 + 𝐵𝜇𝑝 + 𝐶

for some constants 𝐴, 𝐵 and 𝐶, with 𝐴 > 0.

This is an equation of a hyperbola.

30
An Example

Question: A portfolio is to be constructed from 2 assets whose mean

and variance of return rate are summarized in the table below. The
correlation of the return rates of these assets is 𝜌. Write down your
answer to 4 significant figures.
Asset Mean return rate Variance of return rate
A 30% 5.76
B 10% 0.36

In part (ii), (iii) and (iv), we assume ρ = 0.6.

31
An Example
1
(i) Mr Investor is considering investing 2 of his initial capital in asset A and
1
the other 2 in either cash (with 0 risk) or in asset B. his objective is to
minimize investment risk. Find the rage of 𝜌 for which he would prefer
cash to B.

(ii) Find the weight vector and the standard deviation of the portfolio that
has the smallest variance under the condition that portfolio mean is at
least 20%.

(iii) Find the weight vector, mean and the standard deviation of the
portfolio that has the smallest variance under the condition that short
selling is not allowed.

(iv) Find the maximal mean of the portfolio that has the risk (the standard
deviation) no more than 0.7. Please justify your answer.

(v) Find the range of 𝜌 for which the risk of the portfolio is possibly 0.
32
Solution

33
Solution

34
Feasible Sets

In the previous example, the graph of 𝜇𝑝 against 𝜎𝑃 for all portfolio

weight 𝒘 is also known as the feasible set of two given assets.

In general, given any two risky assets 1 and 2, it can be shown that
the feasible set
i. Is a straight line joining 𝜎1 , 𝜇1 and 𝜎2 , 𝜇2 when 𝜌12 = 1
(perfectly positively correlated)
ii. Is a V-shaped graph comprising two straight lines, each joining
the (𝜎, 𝜇) point of one asset to a point with zero portfolio
variance, when 𝜌12 = −1 (perfectly negatively correlated)
iii. is a curve passing through the (𝜎, 𝜇) points of the two assets
when |𝜌12 | < 1

35
Feasible Set
For the case when 0 ≤ 𝛼 ≤ 1 (i.e. short-selling is not allowed), a typical
feasible set for the various cases outlined in the previous slides is depicted
in the diagram below. If short-selling is possible, the feasible set can be
obtained by simply extending the corresponding graph below/beyond the
end points

36
Portfolios of Three or More Assets

When there are 𝑛 (𝑛 ≥ 3) assets, different portfolios corresponding

to different weight vectors, 𝒘 = 𝑤1 , 𝑤2 , … , 𝑤𝑛 𝑇 can be formed.
The feasible set contains all the 𝜎𝑝 , 𝜇𝑝 points.
For the case of three assets, the feasible set is the solid region
enclosed by all the curves formed by portfolios with two assets.

Feasible set for three or more assets 37

Portfolios of Three or More Assets
We have seen in the previous slides that any two assets define a
line/curve btw their (𝜎, 𝜇) points. If we take any portfolio (label it as
asset 4) in the feasible set defined by assets 2 and 3, and combine
it with asset 1, a line/curve btw 1 and 4 will be formed, as shown
below. By taking any pair of assets on or within the boundary
curves formed by assets 1, 2 and 3, a solid region will be traced
out.

38
Portfolios of Three or More Assets

It can be shown that the feasible set, 𝐹 of a portfolio with 𝑛 (𝑛 > 2)

assets has the following properties:
i. For any fixed 𝜇 ∈ 𝑅, 𝜎, 𝜇 𝜖 𝐹 for some 𝜎 > 0
ii. For each 𝜎, 𝜇 𝜖 𝐹, 𝜎′, 𝜇 𝜖 𝐹 for all 𝜎 ′ > 𝜎
iii. For each pair of points 𝜎, 𝜇 and 𝜎′, 𝜇′ in the feasible set 𝐹
and for any λ 𝜖 [0,1], the point λ 𝜎, 𝜇 + (1 − λ) 𝜎′, 𝜇′ lies in the
set 𝐹. We say that 𝐹 is a convex set.
iv. For any fixed 𝜇 ∈ 𝑅, there exists 𝜎 ∗ > 0 s.t.
a) We have 𝜎 ∗ , 𝜇 𝜖 𝐹
b) If 𝜎, 𝜇 𝜖 𝐹, then 𝜎 ∗ < 𝜎
We call the point 𝜎 ∗ , 𝜇 the minimum-variance point with
mean 𝜇

39
Portfolios of Three or More Assets

40
Minimum-Variance Set and
Efficient Frontier

We know from Property (iv) of a feasible set that there is a

minimum-variance point for any portfolio mean. The set of all
minimum-variance points is called the minimum-variance set or
minimum variance frontier. For a given return 𝜇, a risk averse
individual seeks the portfolio with the smallest risk, which is the
minimum-variance portfolio with mean 𝜇.

We will show in the next Chapter that the minimum-variance set is a

hyperbolic curve as shown in the previous diagram. The extreme left
point on this frontier is called the global minimum-variance point.
This corresponds to the portfolio with the smallest risk.

41
Minimum-Variance Set and
Efficient Frontier
If the portfolio risk 𝜎 is given, an individual will desire the portfolio
with the highest return. As shown in the diagram below, this
portfolio is the one that lies on the upper half of the minimum-
variance frontier, called the efficient frontier.

42