Markowitz Portfolio Theory

 Expected returns, variance, covariance and correlation.

 Portfolio return and risk with two assets.
 Short selling.
 Two stock portfolio combination lines.
 Three or more stock portfolios.
 Markowitz bullet of risky assets
 Tangent portfolio, Sharpe ratio and CML.
 Two fund separation theorem.
 Portfolio optimisation using spreadsheets.
Single Period Returns
Total discretely compounding return from time 0 to 1:

𝑝1 − 𝑝0 + 𝑑𝑖𝑣1 𝑝1 + 𝑑𝑖𝑣1
𝑟0−1 = = −1
𝑝0 𝑝0

Where
𝑝0 = price at time zero, or the buy price
𝑝1 = price at time one, or the sell price
𝑑𝑖𝑣1 = dividend cash flow received at time one.

2
Arithmetic Average Returns

Arithmetic average return from time 0 to 𝑛:

∑𝑛𝑖=1(𝑟𝑖 ) 𝑟1 + 𝑟2 + ⋯ + 𝑟𝑛
𝑟̅ 0−𝑛 = =
𝑛 𝑛

3
Risk
The common sense idea of risk is the chance of losing
money.
However, in finance risk tends to be measured as the
deviation of returns around the expected (average) return
because this makes the mathematics more tractable. Note
that this definition of risk means that deviation below and
above the expected return is classified as risk.

4
Measures of Risk: Variance and
Standard Deviation

Variance of returns over n periods:

𝑛 2]
2
∑ 𝑖=1[( 𝑟𝑖 − 𝑟̅ )
𝑣𝑎𝑟(𝑟) = 𝜎 =
𝑛−1

Standard deviation of returns over n periods:

∑𝑛𝑖=1[(𝑟𝑖 − 𝑟̅ )2 ]
𝑠𝑑 (𝑟) = 𝜎 = √𝑣𝑎𝑟(𝑟) = √
𝑛−1
5
Sample and Population Statistics
 Note that the above variance and standard deviation
formulas are sample statistics since we divide by (n-1).
 For population versions of these formulas, divide by
(n) instead of (n-1).
 We almost always use the sample statistic formulas
since we usually work with a sample of time series
data.
 Strictly, the symbol for sample variance and sample
standard deviation should be written as 𝑠 2 and 𝑠, not
𝜎 2 and 𝜎. But in this course we’ll use the latter
notation.

6
Covariance and Correlation over Time
Covariance of returns between stocks A and B over n periods:
∑𝑛𝑖=1[(𝑟𝐴,𝑖 − 𝑟̅𝐴 )(𝑟𝐵,𝑖 − 𝑟̅𝐵 )]
𝑐𝑜𝑣 (𝑟𝐴 , 𝑟𝐵 ) = 𝜎𝐴,𝐵 =
𝑛−1
Correlation coefficient of returns between stocks A and B:
𝑐𝑜𝑣 (𝑟𝐴 , 𝑟𝐵 ) 𝜎𝐴,𝐵
𝑐𝑜𝑟𝑟𝑒𝑙 (𝑟𝐴 , 𝑟𝐵 ) = 𝜌𝐴,𝐵 = =
𝑠𝑑(𝑟𝐴 ). 𝑠𝑑 (𝑟𝐵 ) 𝜎𝐴 . 𝜎𝐵
The correlation coefficient is more intuitively useful since it
will always be between -1 and 1, whereas the covariance could
be anything from +∞ to -∞.

7
Strictly, the correlation coefficient is only defined between two
variables. The correlation between a constant (with zero
standard deviation) and a variable will lead to a division by
zero which is undefined.

8
Correlation: How It Looks
𝝆=𝟏 𝝆=𝟎 𝝆 = −𝟏
Perfectly positive Independent Perfectly negative

9
Diversification and Correlation
Diversification is the reduction of risk by combining assets
in a portfolio.
The amount of diversification achievable is inversely
related to the correlation of returns between assets in a
portfolio.
The higher the correlation between two assets, the less
diversification that’s possible when combining them in a
portfolio, which is a pity. For example, two bank stocks
are likely to have a high correlation with each other (close
to one), so a portfolio of these two similar stocks is
unlikely to reduce risk much at all.
10
Ideally, negative correlations are the best.

11
Calculation Example
Q1) Find the discrete yearly returns of stocks CBA and BHP
from the following price data.
Adjusted Closing Price ($)
Date
CBA BHP
1/1/2007 48.39 25.17
2/1/2008 47.77 36.31
2/1/2009 26.01 30.11
4/1/2010 51.47 38.9
4/1/2011 52.46 44.25

12
To find CBA’s return from 2010 to 2011,
𝑝4/1/2011
𝑟𝐶𝐵𝐴,2010→2011 = −1
𝑝4/1/2010
52.46
= − 1 = 0.0192
51.47
And so on. Here are the complete results:
Return (p.a.)
Date
CBA BHP
1/1/2007
2/1/2008 -0.0128 0.4426
2/1/2009 -0.4555 -0.1708
4/1/2010 0.9789 0.2919
4/1/2011 0.0192 0.1375

13
Q2) Calculate the arithmetic mean, variance and standard
deviation of returns.
𝑟𝐶𝐵𝐴,07→08 + 𝑟𝐶𝐵𝐴,08→09 + 𝑟𝐶𝐵𝐴,09→10 + 𝑟𝐶𝐵𝐴,10→11
𝑟̅ 𝐶𝐵𝐴, =
2007→ 𝑛
2011,𝑝𝑎
−0.0128 − 0.4555 + 0.9789 + 0.0192
= = 0.1324
4
𝑛 2]
2
∑ 𝑖=1[( 𝑟𝑖 − 𝑟̅ )
( )
𝑣𝑎𝑟 𝑟 = 𝜎 =
𝑛−1
(−0.0128 − 0.1324)2 +
(−0.4555 − 0.1324)2 +
(0.9789 − 0.1324)2 +
[ ( )2 ]
2 0.0192 − 0.1324
𝜎𝐶𝐵𝐴,2007→2011,𝑝.𝑎. = = 0.3653
4−1
14
 2
𝜎𝐶𝐵𝐴,2007→2011,𝑝.𝑎. = 0.3653

𝜎𝐶𝐵𝐴,2007→2011,𝑝.𝑎. = √0.3653
= 0.6044
Similarly for BHP. Here are the complete results:
CBA BHP
Return 0.1324 0.1753
Variance 0.3653 0.0687
St. dev. 0.6044 0.2622

15
Q3) Calculate the covariance and correlation of their returns.
∑𝑛𝑖=1[(𝑟𝐴,𝑖 − 𝑟̅𝐴 )(𝑟𝐵,𝑖 − 𝑟̅𝐵 )]
𝑐𝑜𝑣(𝑟𝐴 , 𝑟𝐵 ) = 𝜎𝐴,𝐵 =
𝑛−1
𝜎𝐶𝐵𝐴,𝐵𝐻𝑃,2007→2011,𝑝.𝑎. =
(−0.0128 − 0.1324) × (0.4426 − 0.1753) +
(−0.4555 − 0.1324) × (−0.1708 − 0.1753) +
[ ]
(0.9789 − 0.1324) × (0.2919 − 0.1753) +
(0.0192 − 0.1324) × (0.1375 − 0.1753) +
4−1
𝜎𝐶𝐵𝐴,𝐵𝐻𝑃,2007→2011,𝑝.𝑎. = 0.0892

16
 𝑐𝑜𝑣(𝑟1 , 𝑟2 ) 𝜎1,2
𝑐𝑜𝑟𝑟𝑒𝑙(𝑟1 , 𝑟2 ) = 𝜌1,2 = =
𝑠𝑑 (𝑟1 ). 𝑠𝑑(𝑟2 ) 𝜎1 . 𝜎2
0.0892
𝜌𝐶𝐵𝐴,𝐵𝐻𝑃,2007−2011,𝑝.𝑎. =
0.6044 × 0.2622
= 0.5629
CBA BHP
Covariance 0.0892
Correlation 0.5629

In conclusion, BHP has a higher return and lower risk than

CBA. The correlation is not near one so a fair amount of
diversification is possible.
17
Expected Returns: 𝑬[𝒓𝒊 ] = 𝝁𝒊 = 𝒓̅𝒊
Stock returns are random variables that change all of the time.
The expected return of a stock is what you expect the stock’s
return to be in the future. Usually the best estimate of the
expected return is the historical average of the stock’s random
returns. So expected return is equal to the average return.
The expected return 𝐸[𝑟𝑖 ] is a constant since it doesn’t change.
𝑟𝑖 represents stock 𝑖’s random return. It’s a variable.
𝐸[𝑟𝑖 ] represents stock 𝑖’s expected return. It’s a constant.
𝜇𝑖 is another common notation for 𝐸(𝑟𝑖 ), pronounced ‘mu’.
𝑟̅𝑖 represents the average return, pronounced ‘r-bar’.
18
Graphs of Random and Expected Returns
The difference between 𝑟𝑖 and 𝐸[𝑟𝑖 ] or 𝜇𝑖 or 𝑟̅𝑖 can be seen in
the below graphs. The graph on the left has time on the x-axis.
𝑟𝑖,𝑡 is stock i’s random return over day t, that is from time t-1
to t. The blue line is the average. On the graph on the right,
stock i is plotted on the 𝜇-𝜎 graph as a single point.

19
Stock Choice
For each diagram which is better, A or B?

20
 In the left diagram, A is clearly better than B since it has
more return and less risk.
 In the right diagram, A has a higher return than B, but more
risk. So the best stock will depend on the investor’s
preferences or ‘risk appetite’.
o But investors are not limited to invest in just one. They
could buy a portfolio - some of A and some of B.
o Having more portfolios and more choices is good.
o By mixing stocks overall portfolio risk should be
reduced due to diversification.
o The effect is that the risk of the whole (the portfolio) is
less than the sum of the parts (the stocks).

21
Portfolios and Diversification
Let P be a $100 portfolio with $50 invested in A and $50 in B.
 Therefore P is an equal-weighted portfolio in A and B. So:
𝑥𝐴 = 0.5, 𝑥𝐵 = 0.5
 Return will be exactly halfway at 0.15.
 Standard deviation is likely to be less
than halfway (0.2) due to
diversification - we didn’t “put all of
our eggs in one basket”.
 So P will be somewhere along the
dashed red line. The standard
deviation of P depends on the
correlation between A and B.
22
23
How Correlation Affects Diversification
 𝜌𝐴,𝐵 = −1, lots of
diversification since A
and B move in opposite
directions.
 𝜌𝐴,𝐵 = 0, diversification
since A and B move
independently and will
sometimes cancel each
other out.
 𝜌𝐴,𝐵 = 1, no diversification at all since A and B move
with each other in the same ratio. In this case 𝜎𝑃 = 0.2.
24
Portfolio Return and Variance
Portfolio return for 𝑛 stocks with weights 𝑥:
𝑛

𝑟𝑃 = 𝑥1 . 𝑟1 + 𝑥2 . 𝑟2 + ⋯ + 𝑥𝑛 . 𝑟𝑛 = ∑(𝑥𝑖 . 𝑟𝑖 )
𝑖=1

Note that the weights must sum to one:

𝑥1 + 𝑥2 + ⋯ + 𝑥𝑛 = 1
Portfolio variance for 2 stocks with weights 𝑥1 and 𝑥2 :
𝜎𝑃 2 = 𝑥12 . 𝜎12 + 𝑥22 . 𝜎22 + 2. 𝑥1 . 𝑥2 . 𝜎1,2

25
Portfolio Return and Variance Example
Question: Portfolio P in the
diagram has a weight of 0.5 in A
and 0.5 in B. The correlation
between A and B is 0.05. Find the
return and variance of portfolio P.
Answer:
𝑟𝑃 = 𝑥𝐴 . 𝑟𝐴 + 𝑥𝐵 . 𝑟𝐵
= 0.5 × 0.2 + 0.5 × 0.1
= 0.15

26
Since 𝜌𝐴,𝐵 = 0.05, portfolio variance is:
𝜎𝑃 2 = 𝑥𝐴2 . 𝜎𝐴2 + 𝑥𝐵2 . 𝜎𝐵2 + 2. 𝑥𝐴 . 𝑥𝐵 . 𝜎𝐴,𝐵
and
𝜎𝐴,𝐵 = 𝜌𝐴,𝐵 . 𝜎𝐴 . 𝜎𝐵
So:
𝜎𝑃 2 = 𝑥𝐴2 . 𝜎𝐴2 + 𝑥𝐵2 . 𝜎𝐵2 + 2. 𝑥𝐴 . 𝑥𝐵 . 𝜌𝐴,𝐵 . 𝜎𝐴 . 𝜎𝐵
= 0.52 × 0.252 + 0.52 × 0.152 +
+ 2 × 0.5 × 0.5 × 0.05 × 0.25 × 0.15
𝜎𝑃 2 = 0.0221875
𝜎𝑃 = 0.1490 Notice that the standard deviation of P is
less than both A and B’s standard deviation. This shows
how diversification can lower risk.
27
Portfolio Weights: to be Long or Short
 Portfolio weights in an investment must sum to one:
𝑥1 + 𝑥2 + ⋯ + 𝑥𝑛 = 1
 Individual weights may be negative. This corresponds to
selling or short selling.
 Short selling can be achieved by borrowing a stock and
selling it, and then later buying another of the same stock
and returning it to the stock lender.
 Investors who are:
o ‘short’ must have sold the stock. They make money
when prices fall.
o ‘long’ must have bought the stock. They make money
when prices rise.
28
Combination Lines: Weights and Returns
For a portfolio P invested in only 2 stocks A and B,
 P must lie on the combination
line.
 If P has a positive weight in A
and B (long A and B), its return
must be between 0.1 and 0.2.
 If P has a negative weight in B
(short B), then it must have a
weight of more than one in A
(long A), and a return of more than 0.2.
 Vice versa for a negative weight in A (short A).
29
Calculation Example: Short Selling
Question: An investor starts with $100 of wealth. She short
sells $150 of stock B by borrowing stock B from an investment
bank (paying a small fee which you can ignore). Then she sells
stock B for $150 on the stock exchange. With the $250 that she
now has, she buys $250 of stock A. This all happens at t=0.
Later, at t=1, she will sell stock A and
then buy stock B on the exchange to
give back to the investment bank.
Using the information in the diagram,
what is the expected return of her
portfolio?

30
Answer 1 (using weights, the quick way):
Calculate the weights in each stock:
+250
𝑥𝐴 = = 2.5, we use +$250 since we longed stock A.
100
−150
𝑥𝐵 = = −1.5, we use -$150 since we shorted stock B.
100

Check that the weights sum to one: 𝑥𝐴 + 𝑥𝐵 = 2.5 − 1.5 = 1

To find the portfolio return,
𝜇𝑃 = 𝑥1 𝜇1 + 𝑥2 𝜇2 + ⋯ + 𝑥𝑛 𝜇𝑛
𝜇𝑃 = 𝑥𝐴 𝜇𝐴 + 𝑥𝐵 𝜇𝐵
= 2.5 × 0.2 + −1.5 × 0.1
= 0.35

31
Answer 2 (using dollars, the long way):
Note that returns are expressed per year and the investment is
over one year. Let 𝑃𝑖𝑡 be the price of stock ‘i’ at time ‘t’:
𝑃𝐴0 = 250 (Price of stock A at time 0)
𝑃𝐵0 = 150 (Price of stock B at time 0)
The portfolio price (𝑃𝑃 ) at t = 0 is long 1 stock A (𝑛𝐴 = 1) and
short 1 stock B (𝑛𝐵 = −1):
𝑃𝑝𝑜𝑟𝑡0 = 𝑛𝐴 . 𝑃𝐴0 + 𝑛𝐵 . 𝑃𝐵0
= 1 × 250 − 1 × 150
= 100 which was her wealth at the start.

32
Having a negative number of stock B (𝑛𝐵 = −1) seems strange.
But if you look at the portfolio from the point of view of the
balance sheet, where assets (V=D+E) equals liabilities (D) plus
equity (E), the portfolio price is the equity or net wealth, stock
A is the asset and stock B is the liability since it’s borrowed:
𝐸 = 𝑉 − 𝐷
100 = 250 − 150

33
Let’s find stock A and B’s expected prices 𝑃𝐴1 and 𝑃𝐵1 .
In one year the lady will have to pay back the stock lender a
single stock B. She owes one stock.
Stock B’s price in one year is expected to be:
𝑃𝐵1 = 𝑃𝐵0 (1 + 𝜇𝐵 )1 (Expected price of stock B at time 1)
= $150 × (1 + 0.1)
= $165
So in one year the lady expects to have to pay $165 to buy
stock B from some other third party owner, and give that
newly purchased share B to the stock lending bank.
There might appear to be a problem here because stock B’s
price should grow by the capital return (=total return minus
34
dividend yield) rather than the total return 𝜇𝐵 , since the lady
will have to buy one stock B from somebody and pay its price,
not any dividends.
However, usually short selling contracts specify that the stock
borrower (lady) must pay the dividends back to the stock
lender (bank) when the dividends are paid from company B to
the lady who holds the title to the stock.
Therefore if you regard the ‘prices’ 𝑃𝐵0 and 𝑃𝐵1 instead as
values of the price and dividends, then it’s correct to grow the
value at time zero into the value at time 1 by the total return.
Alternatively, for simplicity you can assume that the stock pays
no dividends so the total return is the capital return.

35
The lady’s stock A asset it’s much simpler, its expected price is:
𝑃𝐴1 = 𝑃𝐴0 (1 + 𝜇𝐴 )1
= $250 × (1 + 0.2)
= $300
The portfolio value at t = 1 is:
𝑃𝑝𝑜𝑟𝑡1 = 𝑛𝐴 . 𝑃𝐴1 + 𝑛𝐵 . 𝑃𝐵1
= 1 × 300 − 1 × 165
= 135
Now we can calculate the portfolio return over the year:
𝑃𝑝𝑜𝑟𝑡1 135
𝜇𝑝𝑜𝑟𝑡0→1 = −1 = − 1 = 0.35
𝑃𝑝𝑜𝑟𝑡0 100
36
Constructing the 2-stock Markowitz Bullet
 In a two-stock world, the combination line is the Markowitz
bullet.
 The line passes through the 2 stocks on the return-
standard deviation graph.

37
Calculation Example
Q) Using the information in the
diagram, find the variance of a
portfolio with a target return of
0.18. Assume 𝜌1,2 = 0.3.
A) There are 2 steps:
1) Find the weight in stock 1 that yields the given return
using the portfolio return equation:
𝑟𝑃 = 𝑥1 . 𝑟1 + 𝑥2 . 𝑟2
Together with the ‘weights sum to one’ equation:
𝑥1 + 𝑥2 = 1, so 𝑥2 = 1 − 𝑥1
So,

38
𝑟𝑃 = 𝑥1 . 𝑟1 + (1 − 𝑥1 ). 𝑟2
0.18 = 𝑥1 × 0.1 + (1 − 𝑥1 ) × 0.2
After solving to find 𝑥1 ,
𝑥1 = 0.2
Therefore
𝑥2 = 1 − 0.2 = 0.8
These weights make sense since the target return of 0.18
is between stock 1 and 2’s returns so we should be long
both stocks, which we are since both weights are positive.

39
2) Now calculate the portfolio variance by substituting the
weights 𝑥1 and 𝑥2 into the portfolio variance equation:

𝜎𝑃 2 = 𝑥12 . 𝜎12 + 𝑥22 . 𝜎22 + 2. 𝑥1 . 𝑥2 . 𝜌1,2 . 𝜎1 . 𝜎2

2 2 2 2
= 0.2 × 0.15 + 0.8 × 0.25 +
2 × 0.2 × 0.8 × 0.3 × 0.15 × 0.25

= 0.0445

𝜎𝑃 = 0.2110

40
Portfolios of 3 or More Stocks
 Portfolios of only 2 stocks are restricted to a
combination line. The combination line comprising
stocks A and B is shown in black.

 After adding stock C, a whole

area of portfolios are
possible. There is a portfolio
possibility ‘cloud’, which is
the grey area in the graph.

41
Constructing the 3+ Stock Markowitz Bullet
This requires a formula for multi-stock portfolio variance.
𝑥1 𝑥2 𝑥3 𝑥4 The grey-shaded part of the table
𝑥1 𝜎1 2 𝜎1,2 𝜎1,3 𝜎1,4 is called the variance-covariance
matrix. It has the variance of each
𝑥2 𝜎2,1 𝜎2 2 𝜎2,3 𝜎2,4 stock along the diagonal, and
𝑥3 𝜎3,1 𝜎3,2 𝜎3 2 𝜎3,4 covariances elsewhere.
𝑥4 𝜎4,1 𝜎4,2 𝜎4,3 𝜎4 2 Note that 𝜎1,2 = 𝜎2,1 and 𝜎1,1 = 𝜎12
Portfolio variance is equal to the sum of each term in the
variance-covariance matrix multiplied by its corresponding
two weights.
𝜎𝑃 2 = 𝑥1 𝑥1 𝜎1,1 + 𝑥1 𝑥2 𝜎1,2 + ⋯ + 𝑥4 𝑥3 𝜎4,3 + 𝑥4 𝑥4 𝜎4,4
42
After collecting like terms and re-arranging, we have the 4-
stock portfolio variance equation:
𝜎𝑃 2 = 𝑥12 𝜎12 + 𝑥22 𝜎22 + 𝑥32 𝜎32 + 𝑥42 𝜎42 +
2𝑥1 𝑥2 𝜎1,2 + 2𝑥1 𝑥3 𝜎1,3 + 2𝑥1 𝑥4 𝜎1,4 +
2𝑥2 𝑥3 𝜎2,3 + 2𝑥2 𝑥4 𝜎2,4 +
2𝑥3 𝑥4 𝜎3,4
Here’s the 3-stock portfolio variance equation:
𝜎𝑃 2 = 𝑥12 𝜎12 + 𝑥22 𝜎22 + 𝑥32 𝜎32
2𝑥1 𝑥2 𝜎1,2 + 2𝑥1 𝑥3 𝜎1,3 +
2𝑥2 𝑥3 𝜎2,3

43
The Markowitz Bullet and the Efficient
Frontier of Risky Assets (excluding rf)
 The Markowitz bullet is the boundary line that contains the
minimum risk portfolios for each return.
 Also called the Minimum Variance Set (MVS) of risky assets
or the portfolio possibility frontier.
 The top half of the Markowitz bullet is called the efficient
frontier.
 The portfolios that comprise the efficient frontier are the
only ones worth investing in because they give the highest
return for a given standard deviation.
 Mathematically, the bullet shaped graph is a hyperbola.

44
45
The Markowitz Bullet with Short Selling
Restrictions
 To short sell stock, investors must locate a stock lender,
usually a broker or investment bank. The broker charges a
fee for lending the stock.
 Sometimes a stock lender cannot be found or the fee is
prohibitive.
 In 2008/09, shorting financial stocks was made illegal for
some time.
 Restrictions on short selling lead to a shrunk Markowitz
bullet.

46
 Without short selling, the highest return possible is that of
the highest return stock. You cannot have a weight of more
than 1 in the highest return stock because weights must
sum to 1 and negative weights in other stocks means short
selling.

 In common sense terms, you can’t

earn a return more than the highest
return stock since you can’t invest
more than your wealth. To over-
invest you must borrow some asset
which is short-selling that asset.

47
The Risk Free Rate (rf) and the Minimum
Variance Set of All Assets
 Risk-free securities have zero standard deviation of
returns. Government bonds (also called Treasuries) are
assumed to be risk free securities.
 The return of the risk-free security is referred to as 𝑟𝑓 . It’s
also used to refer to the security itself.
 When 𝑟𝑓 is included, the new MVS becomes a line from 𝑟𝑓
through the tangency portfolio (T) on the Markowitz bullet.
 𝑟𝑓 is a constant, so it has zero variance, and zero covariance
with other securities.
 𝑟𝑓 = 𝐸(𝑟𝑓 ) = 𝜇𝑟𝑓 , since 𝑟𝑓 is a constant.
48
49
The Tangency Portfolio
 The tangency portfolio T is the only risky portfolio worth
investing in. It is comprised of the stocks A, B and C.
 For any level of risk (standard deviation), the highest
return possible can be achieved by investing in T and 𝑟𝑓 .
 Lines from 𝑟𝑓 through any portfolio
are called Capital Allocation Lines.
 The CAL through T has the best risk-
return trade off. It has the steepest
gradient (rise/run).
 The gradient of the CAL is also called
the Sharpe ratio.

50
Sharpe Ratio
The Sharpe ratio (S) of a stock is the gradient of the line from
𝑟𝑓 through the stock. It is the gradient of the stock’s CAL.
𝜇𝑖 −𝑟𝑓
𝑆𝑖 =
𝜎𝑖

Where 𝑆𝑖 is the Sharpe ratio of stock

‘i’, 𝜇𝑖 is its expected return and 𝜎𝑖 is its
standard deviation.
In the diagram, portfolio T’s Sharpe
ratio is greater than stock B’s since
the CAL through T is steeper.
Therefore portfolio T is preferable to
stock B.
51
The Market Portfolio
 The market portfolio M is the tangency portfolio of all risky
assets.

 The line through M and 𝑟𝑓 is

called the Capital Market Line
(CML).

 The CML has the steepest

gradient, therefore the market
portfolio has the highest Sharpe
ratio.

52
Equation of the Capital Market Line (CML)
𝜇𝑚 −𝑟𝑓
𝜇 = 𝑟𝑓 + 𝜎 ( )
𝜎𝑚

This is easy to see since the:

 y-axis is expected return 𝜇
 x-axis is standard deviation 𝜎
 y-intercept is 𝑟𝑓
𝑟𝑖𝑠𝑒
 gradient between 𝑟𝑓 and M is
𝑟𝑢𝑛
𝜇𝑚 −𝑟𝑓
𝜎𝑚

So in 𝑦 = 𝑚𝑥 + 𝑏 form:
𝜇𝑚 − 𝑟𝑓
𝜇=( ) 𝜎 + 𝑟𝑓
𝜎𝑚
53
Calculation Example
Question: Assume a 3-stock world consisting of A, B and C, as
well as the risk free security. The market portfolio has been
calculated to have weights 1/3 in each of A, B and C. The risk
free rate 𝑟𝑓 = 0.05, the market return is 𝑟𝑚 = 0.3 and the
market’s standard deviation is 𝜎𝑚 = 0.2.
Find the weights in stocks A, B, C and 𝑟𝑓 which makes an
efficient portfolio (P) with a return of 𝑟𝑝 𝑡𝑎𝑟𝑔𝑒𝑡 = 0.1. Also find
this portfolio’s standard deviation 𝜎𝑝 𝑡𝑎𝑟𝑔𝑒𝑡 .
Answer: An efficient portfolio has minimum variance (or st.
dev.) for a given return. All portfolios on the CML are efficient.

54
Therefore we need only consider investing in the market
portfolio (M) and the risk free rate (𝑟𝑓 ).
To find the weights we need to invest in M and 𝑟𝑓 , we will use
the portfolio return equation, with a target portfolio return of
0.1:
𝑟𝑃 = 𝑥1 . 𝑟1 + 𝑥2 . 𝑟2 + ⋯ + 𝑥𝑛 . 𝑟𝑛
0.1 = 𝑥𝑀 × 0.3 + 𝑥𝑟𝑓 × 0.05
Now we’re stuck since we have 2 unknowns (𝑥𝑀 and 𝑥𝑟𝑓 ) and
only one equation so we can’t find either of the weights. But
there is another equation, the ‘sum of the weights equals one’:
𝑥1 + 𝑥2 + ⋯ + 𝑥𝑛 = 1
𝑥𝑀 + 𝑥𝑟𝑓 = 1
55
𝑥𝑟𝑓 = 1 − 𝑥𝑀
Substitute this into the portfolio return equation to get:
0.1 = 𝑥𝑀 × 0.3 + (1 − 𝑥𝑀 ) × 0.05
𝑥𝑀 = 0.2
So, 𝑥𝑟𝑓 = 1 − 0.2 = 0.8
This makes sense since the target
return of 0.1 is closer to 𝑟𝑓 so it
should have a larger weight in 𝑟𝑓
than M.
Since M is 1/3 in each of A, B and C, the weights in A, B and C
are simply 1/3 of the weight in M:

56
 1
𝑥𝐴 = 𝑥𝐵 = 𝑥𝐶 = × 𝑥𝑀
3
1 3
= × 0.2 = = 0.0666667
3 15
To find this efficient portfolio’s standard deviation, we could
use the 2-stock portfolio variance equation with 𝑟𝑓 and M and
the weights we just found, together with the fact that the
covariance of 𝑟𝑀 with 𝑟𝑓 is zero since 𝑟𝑓 is a constant:

𝜎𝑃 2 = 𝑥12 . 𝜎12 + 𝑥22 . 𝜎22 + 2. 𝑥1 . 𝑥2 . 𝜎1,2

57
But another faster method is to use the CML equation instead:
𝜇𝑚 −𝑟𝑓
𝑟𝑃 = ( ) 𝜎𝑃 + 𝑟𝑓
𝜎𝑚

Where 𝑟𝑃 is the return of our

efficient portfolio of 0.1, and 𝜎𝑃 is
the variable we are trying to find.
0.3 − 0.05
0.1 = ( ) 𝜎𝑃 + 0.05
0.2
𝜎𝑃 = 0.04

58
Tobin’s Two Fund Separation Theorem
The Two Fund Separation Theorem says that the investing and
financing decision are separate problems:
 The investment decision: which risky assets comprise the
market portfolio? It is the purely technical problem of
identifying the market portfolio – finding the weights in the
risky assets which give the tangency portfolio.

 The financing decision: how much to invest in 𝑟𝑓 and how

much to invest in the market portfolio M? This depends on
the investor’s risk preferences.

59
60
Time Series Calculations
Question: Using the following data, calculate the discrete
yearly returns of the stocks, as well as the arithmetic average
return, standard deviation, variance, covariance and
correlation of their returns.
Adjusted Closing Price ($)
Date
CBA BHP
4/1/2011 52.46 44.25
4/1/2010 51.47 38.9
2/1/2009 26.01 30.11
2/1/2008 47.77 36.31
1/1/2007 48.39 25.17

61
Time Series Calculations – MS Excel
A B C
1 Date Return (p.a.) Return (p.a.)
2 CBA BHP
3 4/1/2011
4 4/1/2010 0.0192 0.1375
5 2/1/2009 0.9789 0.2919
6 2/1/2008 -0.4555 -0.1708
7 1/1/2007 -0.0128 0.4426
8
9 average =average(B4:B7) =average(C4:C7)
10 var =var(B4:B7) =var(C4:C7)
11 stdev =stdev(B4:B7) =var(C4:C7)^(1/2)
12 covar =covar(B4:B7, C4:C7)*4/(4-1)
13 correl =correl(B4:B7, C4:C7)
14
62
Multi-stock Portfolio Variance
Question: Assume there are 3 stocks with the following
variances and covariances:
𝜎12 = 0.01
𝜎22 = 0.04
𝜎32 = 0.09
𝜎1,2 = 0.01
𝜎1,3 = 0.015
𝜎2,3 = 0.03
Find the variance of an equi-weighted portfolio of these 3
stocks.
63
 A B C D
1 weights =1/3 =1/3 =1/3
2 =B1 0.01 0.01 0.015
3 =C1 =C2 0.04 0.03
4 =D1 =D2 =D3 0.09
5 Sumproduct method
=sumproduct(B2:B4 =sumproduct(C2:C4, =sumproduct(D2:D4
6
,$A$2:$A$4) $A$2:$A$4) ,$A$2:$A$4)
=sumproduct(B6:D6
7 var(P) 0.027777778
,B1:D1)
8 Matrix method
=mmult(mmult(B1:
9 var(P) 0.027777778
D1, B2:D4), A2:A4)
10 Note: you must press ctrl+shift+enter after
11 typing the matrix formula in B9.
12

64
MS Excel: Tangency Portfolio
Question:
Find the tangency portfolio T by
investing in the 3 stocks whose
returns and standard deviations are
given in the below graph. The return
of the risk free rate is also given in
the graph. Additionally, the
correlations between the stocks are
all 0.5.

65
Answer:
Graphically, we are trying to find the
weights in stocks 1, 2 and 3 which
give the tangency portfolio T. We’ll do
this by maximizing the gradient of the
CAL. The gradient is the Sharpe ratio.
The maximization part requires the
MS Excel add-in ‘solver’.
First enter the below cells:

66
 A B C D
1 stock 1 2 3
2 stock return 0.1 0.2 0.3
3 weight in P =1/3 =1/3 =1/3
4
5 0.04 0.02 0.02
6 var-cov matrix =C5 0.04 0.02
7 =D5 =D6 0.04
8
9 rf 0.05
=MMULT(MMULT(B3:D3, B5:D7),
10 var(P)
TRANSPOSE(B3:D3))
11 stdev(P) =B10^0.5
12 return(P) =MMULT(B2:D2, TRANSPOSE(B3:D3))
13 sum of weights =SUM(B3:D3)
14 Sharpe ratio(P) =(B12-B9)/B11
15 Note: you must press ctrl+shift+enter after
16 typing the matrix formulas in B10 and B12.
17

67
Now we have to install the solver add-in.
 Click the ‘File’ button in the top left of MS Excel, and a menu
should pop up.
 Click ‘Options’ in the bottom of the menu, and a window
should pop up.
 Click ‘Add-Ins’ on the left margin of the menu.
 Click on the ‘Solver Add-in’ and then click on the ‘Go’
button. Don’t click OK, that does nothing!
 Another menu will pop up. Make sure ‘Solver Add-in’ is
ticked, then click OK.
 Click on the ‘Data’ tab at the top of the excel window.
 You should now see a ‘Solver’ button on the right.

68
 Click the solver button and fill out the solver menu as
follows:

 What we’re trying to do is maximize the Sharpe ratio (B14)

by changing the weights (B3:D3), while keeping the sum of
the weights equal to one (B13).
 Click solve, and hopefully you get the following results,
which gives us the weights in the tangency portfolio, as
well as its return and standard deviation:
69
 A B C D
1 stock 1 2 3
2 stock return 0.1 0.2 0.3
3 weight in P -0.555570888 0.3333 1.2222
4
5 0.04 0.02 0.02
6 var-cov matrix 0.02 0.04 0.02
7 0.02 0.02 0.04
8
9 rf 0.05
10 var(P) 0.058273088
11 stdev(P) 0.241398194
12 return(P) 0.377781949
13 sum of weights 1
14 Sharpe ratio(P) 1.357847561
15 Note: you must press ctrl+shift+enter after
16 typing the matrix formulas in B10 and B12.
17
Note that portfolio P here is actually the tangency portfolio T.

70