AIN3220

Investment and Risk Analysis

Chapter 1
Chapter 1: Asset Returns, Basic Statistics,
and Matrices
Questions:
• What do we mean by risky assets?
• How can we calculate the expected returns and variances of the risky assets?
Topics in this Chapter:
1.1 Asset Returns.
1.2 Basic Statistics (Review).
1.3 Matrices (Review).

AIN3220 Investment and Risk Analysis 1-1

1.1 Asset Returns
All asset returns in this course will be reported as annual returns unless
otherwise stated.
1.1.1 Asset return

Pt+1 − Pt
rt = .
Pt

Pt asset price at time t. rt annual rate of return.

Pt+1 asset price at time t + 1. t time period (in years).

Example 1.1 The asset price at time t is $100 and at time t + 1 is $120.
What is the asset return?
Answer:
$120 − $100
rt = = 0.20.
$100

AIN3220 Investment and Risk Analysis 1-2

1.2 Basic Statistics (Review)
In reality, the future asset returns are uncertain.
When the asset return is uncertain, we call this asset a risky asset (e.g.
stocks), as shown in Example 1.2 below.
On the other hand, when there is no uncertainty in the return of an asset (e.g.
US Treasury bonds), we call such an asset a risk-free asset.
We model asset returns as random variables in this course. Therefore, it is
important to review the basic statistics of random variables (AMS 1302).
1.2.1 Random variables and probability density functions
Random variables
• The value of a random variable is unknown until it is observed.
• A discrete random variable can take only countably many values
x1, x2, · · · . The outcome of rolling a die or tossing a coin is a discrete
random variable.
• A continuous random variable takes values in an interval (a, b) or over

AIN3220 Investment and Risk Analysis 1-3

 the whole real line.
Probability density functions
• For a discrete random variable X, the probability density function,
p(x), gives the probability that X takes the value x:
p(x) = P(X = x).
• Therefore 0 ≤ p(x) ≤ 1.
• If X can take the values x1, · · · , xM , then
p(x1) + · · · + p(xM ) = 1.
Example 1.2 Assume that the current price of an asset is $100 and its future
price is $120 with probability 43 , $80 with probability 18 , and $60 with proba-
bility 81 . What are the possible values for the asset return and the probability
density function associated with these asset returns?
Answer: Denote r to be the asset return. Then,

120−100 3
 100 = 0.20, with p(0.20) = 4 ,

r = 80−100
100 = −0.20, with p(−0.20) = 1
8,

 60−100 1
100 = −0.40, with p(−0.40) = 8.
AIN3220 Investment and Risk Analysis 1-4
1.2.2 Population statistics
Population mean
• For a discrete random variable X, the population mean, µX , is
M
X
µX = E[X] = p(xi)xi.
i=1

Population variance
2
• For a discrete random variable X, the population variance, σX , is
h i XM
2
σX = Var(X) = E (X − µX )2 = p(xi) (xi − µX )2 .
i=1

Example 1.3 What are µr and σr2 in Example 1.2 ?

Answer:
3 1 1
µr = × (0.20) + × (−0.20) + × (−0.40) = 0.075,
4 8 8
AIN3220 Investment and Risk Analysis 1-5
and
3 1 1
σr2 = × (0.20 − 0.075) + × (−0.20 − 0.075) + × (−0.40 − 0.075)2
2 2
4 8 8
= 0.049375.

AIN3220 Investment and Risk Analysis 1-6

Population covariance
• The population covariance, σX,Y , for two random variables X and Y
is
 
σX,Y = Cov(X, Y ) = E (X − µX ) (Y − µY ) ,

PM
which can be calculated using i=1 p(xi , yi ) (xi − µX ) (yi − µY ).

Population correlation coefficient

• The population correlation coefficient, ρX,Y , for two random vari-
ables X and Y is
σX,Y
ρX,Y = .
σX σ Y

AIN3220 Investment and Risk Analysis 1-7

1.2.3 Sample statistics
Sample mean
• The sample mean, µ̂X , of a random variable X, based on the observations
x1, . . . , xN , where N is the sample size, is
N
1 X
µ̂X = xi .
N i=1

Sample variance
2
• The sample variance, σ̂X , of a random variable X, based on the obser-
vations x1, . . . , xN , is
N
2 1 X
σ̂X = (xi − µ̂X )2.
N − 1 i=1

AIN3220 Investment and Risk Analysis 1-8

Sample covariance
• The sample covariance, σ̂X,Y , of the random variables X and Y , based
on joint observations (x1, y1), . . . , (xN , yN ), is
N
1 X
σ̂X,Y = (xi − µ̂X ) (yi − µ̂Y ) .
N − 1 i=1

Sample correlation coefficient

• The sample correlation coefficient, ρ̂X,Y , for two random variables X
and Y based on joint observations (x1, y1), . . . , (xN , yN ), is
σ̂X,Y
ρ̂X,Y = .
σ̂X σ̂Y

AIN3220 Investment and Risk Analysis 1-9

Example 1.4 Consider the following data for the returns of two risky assets:
Month 1 2 3 4 5
rA 0.04 -0.02 0.08 -0.04 0.04
rB 0.02 0.03 0.06 -0.04 0.08

Compute the following sample statistics: µ̂A, µ̂B , σ̂A2 , σ̂B2 , σ̂A,B , and ρ̂A,B .
Answer:
µ̂A = 0.02, µ̂B = 0.03, σ̂A2 = 0.0024, σ̂B2 = 0.0021,
σ̂A,B = 0.0017, ρ̂A,B = 0.75724.

AIN3220 Investment and Risk Analysis 1-10

1.2.4 Linear combinations of two random variables
• Portfolios are the linear combinations of assets. Because we treat asset
returns as random variables, portfolio returns can be expressed as the linear
combinations of random variables.
• Let X and Y be two random variables, and let Z be the following linear
combination of X and Y :
Z = aX + bY,
then
µZ = aµX + bµY
and
σZ2 = a2σX
2
+ b2σY2 + 2abσX,Y .
• For sample statistics, these equations are the same but with sample means,
variances, and covariances.

AIN3220 Investment and Risk Analysis 1-11

Example 1.5 Consider again the data from Example 1.4 for the two assets
A and B, but now consider a portfolio, P , that consists of 1/2 of an investment
in A and the other 1/2 in B. The returns of the two assets and the portfolio
are:
Month 1 2 3 4 5
rA 0.04 -0.02 0.08 -0.04 0.04
rB 0.02 0.03 0.06 -0.04 0.08
rP 0.03 0.005 0.07 -0.04 0.06

What are µ̂P and σ̂P2 ?

AIN3220 Investment and Risk Analysis 1-12

Answer: We can calculate µ̂P and σ̂P2 in two ways and both should give the
same result.
• Method 1:
We can calculate using the formulas for sample mean and variance,
0.03 + 0.005 + 0.07 − 0.04 + 0.06
µ̂P = = 0.025,
5
(0.03 − 0.025)2 + (0.005 − 0.025)2 + . . .
σ̂P2 = = 0.001975.
5−1
• Method 2:
We can calculate using the formulas for linear combinations of random vari-
ables. Notice that if X = rA and Y = rB , then Z = rP with a = 1/2 and
b = 1/2.
1 1
µ̂P = × 0.02 + × 0.03 = 0.025,
2 2
1 1 1
σ̂P2 = × 0.0024 + × 0.0021 + 2 × × 0.0017 = 0.001975.
4 4 4

AIN3220 Investment and Risk Analysis 1-13

Note: Two methods give the same result. Notice also that portfolio P has
lower risk (a lower variance of returns) than either A or B. By investing in
A and B together, you reduce your overall risk!!
1.2.5 Excel commands for basic statistics

You will need to know how to calculate sample statistics in Excel for the course
assignment.
• Sample mean: AVERAGE( ).
• Sample variance: VAR.S( ).
• Sample standard deviation: STDEV.S( ).
• Sample covariance: COVARIANCE.S( , ).
• Sample correlation coefficient: CORREL( , ).

AIN3220 Investment and Risk Analysis 1-14

1.3 Matrices (Review)
The equations we will use in this course become very messy with more than
two assets. Therefore, we will adopt matrix notation to simplify things.
1.3.1 Matrices
• A matrix A is a rectangular array of numbers:
 
a1,1 · · · a1,n
A =  .. . . . ..  = (ai,j )m×n.
am,1 · · · am,n
The numbers ai,j are called the components of A.
• The dimension of a matrix is given by its number of rows and columns.

- The dimension of A is m × n (it has m rows and n columns).

- A matrix with the same number of rows as columns (m = n) is called
square matrix.

AIN3220 Investment and Risk Analysis 1-15

• A column vector is an m × 1 matrix, i.e. n = 1:
 
x1
x =  ..  .
xm

• A row vector is a 1 × n matrix, i.e. m = 1:

y = (y1 · · · yn).

AIN3220 Investment and Risk Analysis 1-16

Example 1.6 Consider the following stock return data:
E[r] Var(r)
Stock A 0.10 0.06
Stock B 0.14 0.08

with Cov(rA, rB ) = 0.03. Present the expected returns in a column vector and
the variance-covariance information in a square matrix.
Answer: The vector of expected returns is
   
E[rA] 0.10
µ= = .
E[rB ] 0.14
The variance-covariance matrix is
   
Var(rA) Cov(rA, rB ) 0.06 0.03
Ω= = .
Cov(rB , rA) Var(rB ) 0.03 0.08

AIN3220 Investment and Risk Analysis 1-17

1.3.2 Basic matrix algebra

A number of operations can be performed on matrices. You will need to know

the following matrix operations for this course.
Matrix transposition
• Transposition, denoted >, swaps the rows and columns of a matrix. This
converts the m × n matrix A into an n × m matrix, A>, as follows:
 >  
a1,1 · · · a1,n a1,1 · · · am,1
A> =  .. . . . ..  =  .. . . . ..  = (aj,i)n×m.
am,1 · · · am,n a1,n · · · am,n

AIN3220 Investment and Risk Analysis 1-18

Matrix addition and subtraction
• Matrices can be added and subtracted if their dimensions agree, with
both operations performed component-wise. If B is an m × n matrix,
 
b1,1 · · · b1,n
B =  .. . . . ..  = (bi,j )m×n,
bm,1 · · · bm,n
then
 
a1,1 + b1,1 · · · a1,n + b1,n
A+B =  .. ... ..  = (ai,j +bi,j )m×n
am,1 + bm,1 · · · am,n + bm,n

and
 
a1,1 − b1,1 · · · a1,n − b1,n
A−B =  .. ... ..  = (aij −bij )m×n.
am,1 − bm,1 · · · am,n − bm,n

AIN3220 Investment and Risk Analysis 1-19

Example 1.7 Suppose that the matrices A and B are:
 
  −8 4
3 1 4
A= , B =  5 3 .
6 2 5
9 −7

Calculate A − B >.
Answer:
     
3 1 4 −8 5 9 11 −4 −5
A − B> = − = .
6 2 5 4 3 −7 2 −1 12
Question: Can you calculate A − B?

AIN3220 Investment and Risk Analysis 1-20

Matrix multiplication
• Matrices can be multiplied by scalars. If α is a scalar, then
 
αa1,1 · · · αa1,n
αA =  .. ... ..  = (αa )
i,j m×n
αam,1 · · · αam,n

is another m × n matrix.
• Matrices can be multiplied together, but only if the number of columns in
the left-hand matrix equals the number of rows in the right-hand matrix.
• The resultant matrix has the same number of rows as the left-hand matrix
and the same number of columns as the right-hand matrix.
• If A is a m × n matrix and C is a n × p matrix,
 
c1,1 · · · c1,p
C =  .. . . . ..  ,
cn,1 · · · cn,p
then the following m × p product matrix is well-defined:
AIN3220 Investment and Risk Analysis 1-21
  Pn Pn 
k=1 a1,k ck,1 ··· k=1 a1,k ck,p
AC = P .. ... .. .
n Pn
k=1 am,k ck,1 · · · k=1 am,k ck,p

Example 1.8 Suppose that α = 2 and that the matrices A and C are:
 
  −1 −5
3 1 4
A= , C =  3 0 .
6 2 5
4 6
Calculate αA and AC.
Answer:    
3 1 4 6 2 8
αA = 2 = ,
6 2 5 12 4 10
 
  −1 −5  
3 1 4  16 9
AC = 3 0 = .
6 2 5 20 0
4 6
Question: Can you compute CA?

AIN3220 Investment and Risk Analysis 1-22

1.3.3 Diagonal and invertible matrices
Diagonal matrix
Diagonal matrix is an n × n matrix with zeros everywhere, except on its
diagonal.
• The n × n diagonal matrix,
 
1 ··· 0
In =  .. . . . ..  ,
0 ··· 1
is very special. It is called the identity matrix because it acts as an
identity for matrix multiplication:

XIn = X = InX,

for any n × n matrix X.

AIN3220 Investment and Risk Analysis 1-23

Invertible matrix
• An n × n matrix X is said to be invertible if there exists an n × n matrix
Y satisfying
XY = In = Y X.
If so, the matrix Y is necessarily unique and is written as X −1. We call
X −1 the inverse of X.
• Note that not all square matrices are invertible!
• In general, inverting a matrix will require the assistance of a computer.
• A special case: A diagonal matrix whose diagonal entries are all non-zero
can easily be inverted by inverting each element along the diagonal.
• Example 1.9 Show that E = D−1 is an inverse matrix of D.
   
1 0 0 1 0 0
D = 0 2 0  , E = 0 12 0 . (1.1)
0 0 3 0 0 13

AIN3220 Investment and Risk Analysis 1-24

 • Answer We need to show that DE = I3.
    
1 0 0 1 0 0 1 0 0
DE = 0 2 0 0 12 0 = 0 1 0 = I3. (1.2)
0 0 3 0 0 13 0 0 1

Example 1.10 Demonstrate that Y = X −1 is the inverse of X.

   
0.3 0 0.2 6 −4 −4
X =  0 0.2 −0.2 , Y = −4 11 6  .
0.2 −0.2 0.5 −4 6 6

Answer: We need to show that XY = I3.

    
0.3 0 0.2 6 −4 −4 1 0 0
XY =  0 0.2 −0.2 −4 11 6  = 0 1 0 .
0.2 −0.2 0.5 −4 6 6 0 0 1

AIN3220 Investment and Risk Analysis 1-25

1.3.4 Matrix algebra in Excel
You will need to know how to manipulate matrices in Excel for the course
assignment.
Named ranges
• In Excel a vector or a matrix is represented by a range of cells. In the case
of a vector, the range is a column or a row; while a matrix is represented by
a rectangular range of cells.
• It is possible to name the ranges you work with:
1. Select the range.
2. Enter the desired name into the Name Box region (located to the left of
the cell entry area).
Note that range names in Excel are not case-sensitive.
Named matrix operations in Excel
• Suppose mat1 and mat2 are two named ranges representing matrices (or
vectors) with the same dimensions. To add them:
1. Enter the formula =mat1+mat2 in the desired cell.
AIN3220 Investment and Risk Analysis 1-26
 2. Select the range of cells to contain the answer. You will thus need to know
before hand the dimensions of the answer.
3. Simultaneously press CTRL-SHIFT-ENTER.
• To subtract mat2 from mat1, perform the same steps as above, but enter
the formula =mat1-mat2 instead.
• Performing the same steps as above, but entering the formulas =mat1+100
and =mat1*10 will add 100 to every component of mat1 and multiply mat1
by the scalar 10, respectively.
• If the number of columns of mat1 equals the number of rows of mat2, then
the formula =MMULT(mat1,mat2) computes the product of these two ma-
trices.
• The formula =TRANSPOSE(mat1) will transpose mat1.
• If mat1 is invertible, then its inverse is computed by entering the formula
=MINVERSE(mat1).

AIN3220 Investment and Risk Analysis 1-27

Cell-based matrix operations in Excel
• Suppose A1:B2 and A4:B5 are cell ranges representing two 2x2 matrices.
To add them:
1. Enter the formula =A1:B2+A4:B5 in the desired cell, A7 for example. The
formula can be entered directly by typing ranges or by using the mouse to
select the cell ranges.
2. Select the range of cells to contain the answer. In this example, because
the answer is a 2x2 matrix, select A7:B8.
3. Simultaneously press CTRL-SHIFT-ENTER.
• All of the other operations are the same as with the Named matrices, re-
placing the matrix name with the appropriate cell range.

AIN3220 Investment and Risk Analysis 1-28