15.

401 Recitation

6: Portfolio Choice
Learning Objectives

 R i off Concepts
Review C t
o Portfolio basics
o Efficient frontier
o Capital market line
 Examples
o XYZ
o Diversification
o Sharpe ratio
o Efficient frontier

2010 / Yichuan Liu 2

Review: portfolio basics

 A portfolio is a collection of N assets (A1, A2, …, AN)

with weights (w1, w2, …, wN) that satisfy

N
o w
i1
i 1

 Each asset Ai has the following characteristics:

o Return: ~ri (random variable)
o Mean return: ri
ret rn: i ,  i
 2
o Variance and std.
std de
dev. of return
o Covariance with Aj:  ij

2010 / Yichuan Liu 3

Review: portfolio basics
 The return of a portfolio is
N
rp   wi ~

~ ri
i 1
 The mean/expected return of a portfolio is

E rp   rp   wi ri
N

i 1
 The variance of a portfolio is
N N
 p2   wi w j ij ;  p   p2
i1 j1

 Note:  ii   i2 ;  ij   ij i j
2010 / Yichuan Liu 4
Example 1: XYZ

Variance Covariance
Variance‐Covariance
E(r)
X Y Z
X 15% 0.090 0.125 0.144
Y 10% 0.040 ‐0.036
Z 20% 0.625

 Whatt is
Wh i th
the expected
t d return
t and
d variance
i off a
portfolio of …
a. (X, Y) with weights (0.4, 0.6)?
b. (X, Y, Z) with weights (0.2, 0.5, 0.3)?
c. (X, Y, Z) with weights (1/3, 1/3, 1/3)?

2010 / Yichuan Liu 5

Example 1: XYZ
 Answer:

p  
a. E r  12%;  2  0.08880;   29.80%
p p

b. E r   14%; 
p
2
p  0.10133;  p  31.83%

c. E r   15%; 
p
2
p  0.13567;  p  36.83%

2010 / Yichuan Liu 6

Example 1: XYZ
 What is the minimum possible variance of a portfolio

with only Y and Z?

Variance‐Covariance
E(r)
X Y Z
X 15% 0 090
0.090 0 125
0.125 0 144
0.144
Y 10% 0.040 ‐0.036
Z 20% 0.625

2010 / Yichuan Liu 7

Example 1: XYZ
 Answer:

Let (w, 1–w ) be the weights for (Y, Z), then

arg min w  0.04
w
2 w1
04  2w
2
1 w 0.036
0 036   1
1 w  0.625
2
0 625 
 First‐order condition:

2w  0.04  21 2w 0.036   21 w  0.625  0

w*  0.8969

 The minimum variance portfolio is

 0.8969,0.1 031
0.8969,0.1031
2010 / Yichuan Liu 8
Example 2: diversification
 Supp
ppose that your portfolio consists of N eq
quallyy
weighted identical assets in the market, each of
which has the following properties:
o Mean = 15%
o Std dev = 20%
o Covariance with any other asset = 0.01
 What is the expected return and std dev of return of
your portfolio if…
o N = 2?
o N = 5?
o N = 10?
o N = ∞?
2010 / Yichuan Liu 9
Example 2: diversification
 Answer:
o Expected return
 
N
1
E r    00.15
p 15  0.15
15

i1 N

o Variance
 0.2 2 
 rp    2   2  N  2   N N  1 2
N
0.2 2 N 0.01 0.01
i1 N i1 j i N  N  N
0.04  1 0.03
  1 0.01  0.01
N  N N

2010 / Yichuan Liu 10

Example 2: diversification
 Answer:
o N = 2:
E rp   15%;  p2  0.0250;  p  15.81%
o N = 5:
E rp   15%;  p2  0.0160
0160;;  p  12.65%
12 65%
o N = 10:
E rp   15%
15%;  p2  0 0130  p  11.40%
0.0130; 11 40%
o N = ∞:
E rp   15%;  p2  0.0100;  p  10.00%
2010 / Yichuan Liu 11
Example 2: diversification

20%

18%
Standard Devviation of Return

16%

14%

12%

10%

8%

6%

4%

2%

0%
0 5 10 15 20 25 30

2010 / Yichuan Liu N 12

Review: diversification

20%

18%
Return

16% Idiosyncratic risk can be diversified away;

investors are not compensated for such risk.
risk
Standard Devviation of R

14%

12%

10%

8%

6% Systematic risk cannot be diversified away;

4% investors are compensated
p with higher
g
2%
expected returns.
0%
0 5 10 15 20 25 30

2010 / Yichuan Liu N 13

Review: efficient frontier

 Given two assets, we can form portfolios with

weights (w, 1–w). As we vary w, we can plot the path
of the mean return and standard deviation of
return of the resulting portfolio.
 The shape of the path depends on the correlation
b t
between the
th two assets.t
 When the correlation is low, a large portion of asset

return variation comes from idiosyncratic risk that

can be diversified away.

2010 / Yichuan Liu 14

Review: efficient frontier
 ρ=1 14

perfectly correlated 13
E

no risk reduction potential 12

p = -1
 ‐1 < ρ < 1 11

Expected Return (%)

p=1

imperfectly correlated 10

p=0
p = .30

some risk reduction potential 9

 ρ = ‐1
1
8
D

perfectly negatively correlated 7

most risk reduction potential 6

0 2 4
6 8 10 12 14 16 18 20
Standard Deviation (%)

Image by MIT OpenCourseWare.

2010 / Yichuan Liu 15

Review: efficient frontier
 We can repeat the previous exercise for N assets:

E(r)

Efficient Frontier

Individual Assets
Global Minimum
Variance Portfolio
Minimum-Variance Frontier

2010 / Yichuan Liu Image by MIT OpenCourseWare. 16

Review: efficient frontier
 The efficient frontier can be described by a function
function

σ*(rp), which minimizes the portfolio std dev given

an expected return:
N
 wi  1
 * rp   min
N N

wi   w w  i j ijj s.t.  iN1

i 1 j 1  wi ri  rp
 i 1
 Analytical solution for σ
σ*((rp) is possible but difficult
to derive.

2010 / Yichuan Liu 17

Review: capital market line
 Efficient frontier + risk
risk‐free
free asset = CML

Market Portfolio
Expected Return

Risk-Free
Asset

Efficient Frontier

Risk

Image by MIT OpenCourseWare.

2010 / Yichuan Liu 18
Example 3: Sharpe ratio
 The Sharp
pe ratio measures the reward‐risk tradeoff of an
asset or a portfolio. It is defined as
r  rf
S
S 

 The higher Sharpe ratio, the more desirable an asset / a
portfolio is. Suppose
pp rf = 5%.
5 What is the portfolio of (A,, B))
with the highest Sharpe ratio?
COV‐VAR
E(r)
A B
A 15% 0.090 0.015
B 10% 0.040
4

2010 / Yichuan Liu 19

Example 3: Sharpe ratio
 Answer:
wrA  1 wrB  rf
max S p  max
w   2 w11 w AB  11 w  B2
w w 2 2 2
A

 Method 1: grid search

1. S
Sett up a grid
id f

for w, e.g., w = 0, 0.1, 0.2, …, 1.0

The finer the grid, the more accurate the result

2. Calculate the Sharpe ratio for each w

3. Find the maximum Sharpe ratio.

2010 / Yichuan Liu 20

Example 3: Sharpe ratio

 Method 1: grid search

w 1–w rp – rf σp Sp
0 1 0.0500 0.2000 0.2500
0.1 0.9 0.0550 0.1897
8 0.2899
8
0.2 0.8 0.0600 0.1844 0.3254
0.3 0.7 0.0650 0.1844 0.3525
0.4 0.6 0.0700 0.1897 0.3689
0.5 0.5 0.0750 0.2000 0.3750
0.6 0.4 0.0800 0.2145 0.3730
0.7 0.3 0.0850 0.2324 0.3658
0.8 0.2 0.0900 0.2530 0.3558
0.9 0.1 0.0950 0.2757 0.3446
1 0 0.1000 0.3000 0.3333
2010 / Yichuan Liu 21
Example 3: Sharpe ratio

0.40

0.38

0.36
Maximum:
0 34
0.34
w* = 0.52
0.32 E(r) = 12.60%
σ = 20.26%
0.30
Max S = 0.3752
0 3752
0.28

0.26

0.24

0.22

0.20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

2010 / Yichuan Liu 22

Example 3: Sharpe ratio

 Method 2: Excel Solver

A B C D E Solver
1 E(r) Asset A Asset B
Set Target Cellll:
2 =B3 =B4
$E$7
3 Asset A 0.15 0.09 0.015
4 Asset B =1 –B3
B3 01
0.1 0 015
0.015 0 04
0.04 Equal To:
Max
5
6 rp – rf σp S By Changing Cell:
7 =f =g =C7/D7 $B$3
$B$

f: SUMPRODUCT(B3:B4, C3:C4) – 0.05

g: SQRT(B3*D2*D3+B3*E2*E3+B4*D2*D4+B4*E2*E4)

2010 / Yichuan Liu 23

Example 3: Sharpe ratio

 Method 2: Excel Solver

A B C D E
1 E(r)
() Asset A Asset B
2 0.52 0.48
3 Asset A 0.52 0.15 0.09 0.015
4 A tB
Asset 0.48
8 0.1 0.015 0.04
5
6 rp – rf σp S
7 0.076 0.202583 0.375154

2010 / Yichuan Liu 25

Example 3: Sharpe ratio
 Method 3: analytical solution
o Full derivation:
rA  rB  p2  1 2
  2w
 12
 21 2w AB  21 w B2 rp  rf  
1

S
2
 p 2
A
 2
w  
1
2 2
p

rA  rB w2 A2  2w1 w AB  1 w2  B2  w A2  1 2w AB  1 w B2 wrA  1 wrB  rf 

 p2
0
   
0  rA  rB  w2 A2  2w1 w AB  1 w  B2  w A2  1 2w AB  1 w B2 wrA  1 wrB  rf 
2

 r  r w 
 2w1 w  1 w   1 2w  1 w wr  r   r  w  rf 
2 2 2 2 2 2
A B A AB B A AB B A B B

 r  r w  1 w   w  1 2w  1 w r  r 

A B AB
2
B
2
A AB
2
B B f

 r  r      r  r  r  r        2   r  r w

A B
2
B AB
2
B B f A B
2
B AB
2
A AB
2
B B f

 r  r   r  r   r  r      r  r    w

A f
2
B B f AB A f
2
B AB B f
2
A AB

w* 
r  r   r  r 
A f
2
B B f AB

r  r      r  r    
A f
22

B AB B f
2

A AB

 0.52
2010 / Yichuan Liu 26
Example 3: Sharpe ratio
 Method 3: analytical solution
o Result only:
The general solution for the 2‐asset Sharpe ratio
maximization
i i ti problem
bl i
is

w* 
r  r   r  r 
A f
2
B B f AB

r  r       r  r  
A f
2
B AB B f
2
A   AB 

2010 / Yichuan Liu 27

Example 4: efficient frontier
 Given the risky assets A and B in the previous

question, what is the efficient frontier?
COV VAR
COV‐VAR
E(r)
A B
A 15% 0.090 0.015
B 10% 0.040

 Given 5% risk
 risk‐free
free rate,
rate what is the capital market
line?

2010 / Yichuan Liu 28

Example 4: efficient frontier

 Table from the previous question:

w 1–w rp σp
0 1 0.1000 0.2000

0.1 0.9 0.1050 0.18

897

0.2 0.8 0.1100 0.1844

0.3 0.7 0.1150 0.1844

0.4 0.6 0.1200 0.1897

0.5 0.5 0.1250 0.2000

0.6 0.4 0.1300 0.2145

0.7 0.3 0.1350 0.2324

0.8 0.2 0.1400 0.2530

0.9 0.1 0.1450 0.2757

1 0 0.1500 0.3000

2010 / Yichuan Liu 29

Example 4: efficient frontier

 Scatter plot of (rp, σp) pairs:

0.16
Efficient frontier
0.14

0.12

0.10 Inefficient portion

of the frontier
0.08

0.06

0.04
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

2010 / Yichuan Liu 30

Example 4: efficient frontier

 Capital market line:

0.16
Tangency portfolio:
w = 0.52
0.14
E(r) = 12.60%
σ = 20.26%
0.12 Max S = 0.3752

0.10

0.08

CML is the line passing through (0, 0.05) and

0.06
tangent to the efficient frontier.

0.04
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

2010 / Yichuan Liu 31

Example 4: efficient frontier
 The moral of the story:
o The CML is tangent to the efficient frontier at the

tangency portfolio.

o The
Th tangency
t portfolio
tf li is
i th
the portfolio
tf li off risky
i k assets
t th
thatt
maximizes the Sharpe ratio.
o The slope of the CML is the maximum Sharpe ratio.
o Rational investors always hold a combination of the
tangency portfolio and the risk‐free asset. The
proportion depends on investors’
investors risk preferences.
preferences.

2010 / Yichuan Liu 32

Sneak Peak: CAPM
 The tangency portfolio is the market portfolio portfolio..
 An asset’s systematic risk is measured by beta, which
is defined as the correlation of its return and the
market return, normalized by the variance of market
return :
 im
i  2
m
 Since investors are only y comp pensated for sy ystematic

risk, asset return is an increasing function of beta:

ri   rf   i ~
E ~ ri  rf 

2010 / Yichuan Liu 33

MIT OpenCourseWare
http://ocw.mit.edu

15.401 Finance Theory I

Fall 2008

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.