Quantitative Methods

Pre-course Reading
 2
Course Contents

A: Foundations
B: Probability & Statistics
C: Calculus

Course text: PRM Handbook Volume II

Topic A: Foundations
A1 Symbols and Rules
A2 Sequences and Series
A3 Exponents and Logarithms
A4 Equations and Inequalities
A5 Functions and Graphs
A6 Compounding Interest
 4
A.1 Symbols and Rules
Expressions
• Expressions contain:
o letters (x, y, z, a, b, c, α, β, γ etc.) which represent variables or
constants, and
o operators (such as +, ,  etc)
• Usually we reserve x, y, z etc for variables (i.e. there are a range of
possible values for x)
• Other letters, like a, b, c, α, β, γ are used for constants (i.e. they take a
specific value)
• Sometimes expressions are accompanied by information about the
domain (i.e. possible values) of the variable(s)
• For instance, a general quadratic expression is:
ax2 + bx + c,  x  R
 5
Mathematical Shorthand

 ‘for all’  ‘implies’

 ‘such that’  ‘if and only if’
 ‘there exists’  ‘identically equal to’
 ‘unique’ = ‘equal to’
 ‘therefore’  ‘not equal to’
 ‘is distributed as’  ‘approximately equal to’
 ‘infinity’ < ‘ less than’
 ‘contained in’ > ‘greater than’
 ‘not contained in’  ‘less than or equal to’
 ‘plus or minus’  ‘greater than or equal to’
 6
Numbers
• Natural numbers (N) {1, 2, 3, 4, 5, …}
• Integers (Z) {…, –3, –2, –1, 0, 1, 2, 3, …}
• Rationals (Q) (or ‘fractions’). They each consist of a pair of
integers, the numerator and the denominator
• Irrationals These numbers cannot be expressed as the ratio
of two integers. Examples of irrationals include π, e, and
numbers such as 0.101001000100001….
• Reals (R) These consist of the rational and the irrational
numbers
• Complex numbers (C) These will not concern us on this
course
 7
Operations

• We are used to to the operations of addition, subtraction,

multiplication and division
• Much time is spent in schools in dealing with operations on
rational numbers, e.g.
a p aq + bp a p ap
+ = and  =
b q bq b q bq

• Other rules are given in the Handbook, p17.

• From these rules we can deduce such results as
a – (–b) = a + b and (–a)  (–b) = a  b
 8
Conventions and Brackets
• Brackets (or parentheses) are used to define the order of
operations
• Sometimes they are omitted in common expressions
• For instance, it is the convention (i.e. common practice) to
write down a linear expression in the form 2x + 3
(understanding that 2x is itself shorthand for 2  x)
• It is convention that this is shorthand for (2x) + 3, rather
than 2(x + 3)
• The acronym ‘BIDMAS’ gives the order in which to apply
operations (e.g. in Excel) unless brackets indicate otherwise
− Brackets, Indices (i.e. powers), Divisions, Multiplications,
Additions, Subtractions
 9
Greek Symbols
Lower- Upper-case Pronunciation Examples in finance
case letter letter
α Α Alpha Regression intercept
β Β Beta Regression slope – systematic risk
γ Γ Gamma Sensitivity measurement for options
δ Δ Delta Sensitivity measurement for options
ε Ε Epsilon Random error or disturbance for regressions
θ Θ Theta Sensitivity measurement for options
κ Κ Kappa Kurtosis
μ Μ Mu Expected return
ν Ν Nu Integer constant, e.g. degrees of freedom
π Π Pi Circle constant – symbol for a product
ρ Ρ Rho Correlation coefficient
σ Σ Sigma Standard deviation – symbol for a sum
τ Τ Tau Time to maturity – maturity date
φ Φ Phi Normal distribution
χ Χ Chi Statistical distribution for testing ‘fit’
 10
Definition of a Function
• A function has the property that, for any value of the input(s) there is a
unique value output
• It is defined by:
Its domain (i.e. the set of elements on which it operates).
• For instance the domain of the logarithmic function is the positive
real numbers. The domain is often not explicitly stated since it can
often be inferred from the context.
Together with its action (i.e. what it does).
• This is usually specified by a rule, like f(x) = 3x + 5 but for finite
and small domains it may be specified by a list, e.g.:
x 1 2 3 4
f(x) 0.1 0.3 0.5 0.1
[The above list is an example where x is a ‘random variable’ and
f(x) is its probability density function – see Topic E]
 11
Types of Functions

• Linear functions, e.g. f(x) = 3x + 5 ( x  R)

• Quadratic functions, e.g. f(x) = 3x2 + 5x – 1 ( x  R)
• Polynomial functions e.g. f(x) = 3x5 + 5x3 – x – 1 ( x  R)
• Special functions: exp(x) and ln(x) are the only ones
covered in this course [e.g. no ‘trigonometric’ functions]
• Functions of several variables, e.g: f(x, y) = x2 + 5ln(y)
( x  R and  y > 0) is a non-linear function of two
variables, x and y
 12
Graphs

• The graph of a function f is obtained by plotting points

defined by all values of x and their associated images
f(x): y-axis

2
(x, y)
y
1

x-axis
x
−2 −1 1 2 3

−1

For the graph we write a function like f(x) = 3ln(x) + 5 as y = 3ln(x) + 5

x is called the independent variable and y is called the dependent variable
 13
Linear Functions

• Consider y = 3x + 5
• The graph is the set of (x, y) such that the y coordinate is 3
times the x coordinate plus 5
• For instance, the point (–3, –4) lies on the line graph

14
12
10
8
6
4
2
0
-3 -2 -1 0 1 2 3
-2
-4
 14
Coefficients vs Parameters
• In the line graph y = 3x + 5, 3 and 5 were coefficients: 3 was the
coefficient of the variable x and 5 was the constant coefficient
• More generally we can write any linear function as
f(x) = m x + c
where
o m is the gradient parameter, i.e. the slope of the line and
o c is the intercept parameter, i.e. the point where the line cuts
the y-axis
• The letters m and c indicate constants that we can change as
we ‘fit’ (i.e. estimate, or calibrate) the function to some data
• For this reason they are called parameters rather than
coefficients
 15
Quadratic Functions
• A quadratic function takes the form:
f(x) = ax2 + bx + c,  x  R
• Graph of y = ax2 + bx + c has parabolic shape

a > 0 ‘’ shape

a < 0 ‘’ shape
b  horizontal location
b > 0  shifted left
b < 0  shifted right
c  vertical location
Topic A: Foundations
A1 Symbols and Rules
A2 Sequences and Series
A3 Exponents and Logarithms
A4 Equations and Inequalities
A5 Functions and Graphs
A6 Compounding Interest
 17
A.2 Sequences and Series
Sequences
• A sequence is also called a progression, and there are two
progressions of importance in finance:
• The Arithmetic Progression: a, a + d, a + 2d, …
This has a first term which is usually denoted by the letter a.
Subsequent terms are derived by repeatedly adding on a fixed
number d, known as the common difference
• The Geometric Progression: a, ar, ar 2, …
This has a first term also usually denoted by a. Subsequent terms
are derived by repeatedly multiplying by a fixed number r, known
as the common ratio
 18
Series

• A series is the accumulated sum of

a sequence
• The conventional notation is:
Sn = the sum of the first n
terms
• And for an infinite series:
S∞ = the limit of the sum to n
terms as n increases
(if that limit exists)
 19
Special Series
Sum of Arithmetic Progression

n
Sn = a + ( a + d ) + ( a + 2 d ) + ... + ( a + ( n − 1)d ) ) = ( 2a + ( n − 1 ) d )
2

Sum of Geometric Progression

a (1 − rn )
Sn = if r < 1;
1−r
a ( rn − 1)
Sn = if r > 1;
r −1
a
S = , if -1 < r < 1 (i.e. if r < 1)
1−r
Topic A: Foundations
A1 Symbols and Rules
A2 Sequences and Series
A3 Exponents and Logarithms
A4 Equations and Inequalities
A5 Functions and Graphs
A6 Compounding Interest
 21
A.3 Exponents and Logarithms
Exponents are also called ‘powers’ or ‘indices’
• Rules
Rule 1: a x a y = a x + y Example: 2 32 4 = 2 7
ax 2 3
Rule 2: y = a x − y Example: 2 3  2 4 = 4 = 2 −1
a 2
Rule 3: a( )x y
=a xy
Example: 2 ( )
3 4
= 2 12

• Consequences include:

a0 = 1
−1 1 −n 1
a = ; a = n
a a
a1 /2 = a ; a1 / n = n a
 22
Inverse Function

• We say that a function g is the inverse of a function f if

g( f (x) ) = x x in the domain

• Then the graph of g is the reflection of f in the 45o line, i.e. the line
y=x
 23
Logarithms

• The logarithm function is the inverse of the exponent

function:

loga ( a y ) = y
or, put another way: loga ( x ) = y  x = a y

• Our rules for exponents translate to rules of logs:

Rule 1: loga ( xy ) = loga ( x ) + loga ( y )

Rule 2: loga ( x / y ) = loga ( x ) − loga ( y )
Rule 3: loga ( x y ) = y loga ( x )
 24
The Exponential Function
Denoted ex, or exp(x)
• e is the irrational number
2.71828182845904523536028747135266249775724709369995957496696762….

20

15

y
10

0
-2 -1 0 1 2 3
x
 25
The Natural Logarithm

• The natural logarithm is the function loge(x) which is

usually denoted ln(x)
• So by definition ln(ex) = x and exp(ln(x)) = x
• In other words, ln (x) is the inverse function of exp(x)
• Hence its graph is the reflection of y = ex in the line y = x
y = ex
y
y=x

y = ln(x)
1

x
1
 26
Expansions of exp(x) and ln(x)

x2 x 3 x 4
exp( x) = x + + + + ...
2! 3! 4!
x2 x 3 x 4
ln ( 1 + x ) = x − + − + ... provided − 1  x  1
2 3 4

• These are very useful for approximations

• For instance, for small values of x :

ln ( 1 + x )  x
Topic A: Foundations
A1 Symbols and Rules
A2 Sequences and Series
A3 Exponents and Logarithms
A4 Equations and Inequalities
A5 Functions and Graphs
A6 Compounding Interest
 28
A.4 Equations and Inequalities
• The capital asset pricing model results in a function
Rs = α + β Ri
where the expected excess return on a security Rs is
linearly dependent on the market portfolio excess
return Ri
• The excess return is the return in excess of the ‘risk-
free’ rate, i.e. the return from holding a risk-less asset
• The coefficient of the market excess return is called the
market beta of the security
• The market-independent return is called the alpha of
the security
• The may be estimated using regression
 29
Example A.1

• Suppose the CAPM predicts Rs = 0.007 + 1.1Ri

• What is the market return if the security return is 8% and
the risk-free interest rate is 4%?
• To answer this set 0.08 – 0.4 = 0.007 + 1.1Ri and solve for Ri
as follows:
o First subtract 0.007 from both sides:
0.033 = 1.1Ri
o Then divide both sides by 1.1:
0.033/1.1 = Ri i.e. Ri = 0.03, or 3%
• The market excess return is 3%
• So the market return is 3% + 4% = 7%
 30
Inequalities
• Linear inequalities are handled in exactly the same way as
linear equalities,
• Except when both sides are multiplied or divided by a
negative number, the direction of the inequality must be
changed.
• Thus, to solve 3 − 2x < 9
o Subtract 3 from both sides: −2x < 6
o Divide both sides by −2
o And reverse the direction of the inequality:
• The result is: x > −3
 31
Inequalities
• Be careful to specify the entire domain
• For instance:
x2 > 9  x > 3 or x < -3
x2 > 9 and x < 0  x < -3
0 < x2 < 9  -3 < x < 3
0 < x2 < 9 and x > 0  0 < x < 3
 32
Systems of Linear Equations
• Consider the solution of two linear equations in two
unknowns:
a1 x + b1 y + c1 = 0 (1st equation)
a2 x + b2 y + c2 = 0 (2nd equation)
where x and y are variables and a1, a2, b1, b2 ,c1 and c2 are
constants
• Values must be found for x and y to satisfy both equations
at the same time
Solution
• One of the equations is solved in terms of one of the
variables and the result is substituted in the other equation
 33
Example A.2

–2x + 4y – 3 = 0 (1st equation)

3x + 2y – 6 = 0 (2nd equation)
Step 1 Solve the first equation in terms of x :
x = (4y – 3)/2 = 2y – 1.5
Step 2 Substitute in the second equation :
3x + 2y – 6 = 0  3(2y − 1.5) + 2y – 6 = 0  8y − 10.5 = 0
 y = 1.3125
Step 3 Determine the value of x :
x = 2y − 1.5 = 21.3125 − 1.5 = 1.125
 34
Quadratic Equations
• Which values of x satisfy the equation
a x2 + b x + c = 0 ?
• A quadratic equation has at most two ‘distinct’ (i.e.
different) solutions, also called the roots of the equation, i.e.
the points where the graph intersects the x-axis
• These can be found (if they exist in R)
o By simple factorisation: e.g. 2x2 + 5x – 3 = (2x – 1)(x + 3)
so the roots are x = ½ and x = -3
o Or using the general formula : −b  b2 − 4 ac
x=
2a
• If b2 < 4ac then there are no real roots
• If b2 = 4ac the two roots are the same
 35
Example A.3
• The roots of x2 + 3x − 4 = 0 are x = 1 and x = −4.
Topic A: Foundations
A1 Symbols and Rules
A2 Sequences and Series
A3 Exponents and Logarithms
A4 Equations and Inequalities
A5 Functions and Graphs
A6 Compounding Interest
 37
A.6 Compounding Interest
Discrete Compounding Factors
• Given a discretely compounded interest rate R the
associated discrete compounding factor is 1 + R
• It is called a compounding factor because multiplying by
1 + R adds on R% interest
• For instance, if the interest rate is 5% then the associated
compounding factor is 1.05:
• So £50 grows into £52.50 if interest of 5% is added, because:
50  1.05 = 52.5
 38
A Limit
Consider 100$ in a savings account with a constant interest rate of
4% per annum (p.a.) and suppose interest is credited to the account
as follows:
1. 4% at the end of each year
The value in the account in 3 years is
100 × 1.043 = 112.486$
2. 2% every 6 months
The value in the account in 3 years is
100 × 1.026 = 112.616$
3. 1% every quarter
The value in the account in 3 years is
100 × 1.0112 = 112.683$
4. 1/3% every month
The value in the account in 3 years is
100 × 1.0033336 = 112.727$
 39
Continuous Compounding
• As the number of times interest is compounded grows, the
value of the 100$ investment increases, but it tends
towards a limit
• This limit is the continuously compounded value of the
investment. It is given by the formula

Continuously compounded value = P exp ( rT )

where
P is the amount invested (called the principal amount),
T is the number of years it is invested and
r denotes the continuously compounded interest rate
 40
Discrete Compounding
• Continuously compounded interest rates are
tractable for theoretical problems but in practice,
interest is compounded at discrete intervals
• If a principal P is invested for T years at an annual
interest rate R :
M
 R
Discretely compounded value = P  1 + 
 N
where
N is the number of compounding intervals per annum,
M is the total number of compounding periods
 41
Example A.4
• If 2$ is invested at a 5% continuously compounded rate, find
its value after 6 months
Solution
2e(0.050.5) = 2e0.025 = 2.05063 $

• If 2$ is invested at a 5% discretely compounded rate, find its

value after 6 months
Solution
6
 0.05 
21 +  = 2.05052 $
 12 
 42
Discrete Rate → Continuous Rate
• Market rates are discrete
• Each market has different conventions
• For this reason it is common practice to convert discrete
rates to continuous rates
• What is the continuous rate r that is equivalent to a discrete
rate R, i.e. both rates give the same compounded sum after a
time period of T years?
• Suppose the discrete rate has compounding frequency N, i.e.
N is the number of times the rate is compounded per year
• Set exp(rT) = (1 + R/N)NT
• Taking logs gives:
r = N ln(1 + R/N)
 43
Return on an Investment
Continuous time notation
• Denote by Y(t ) the value of an investment at time t
• The return on the investment between times t1 and t2
depends on whether we assume discrete or continuous
compounding of interest: the return is
Y ( t2 ) − Y ( t1 )
under discrete compounding
Y ( t1 )
or
 Y ( t2 ) 
ln 
 Y ( t ) 
under continuous compounding
 1 

• Here we have used continuous time notation

 44
Return on an Investment
Discrete time notation
• Denote by Pt the value of an investment at time t
• The return on the investment between times t1 and t2
depends on whether we assume discrete or continuous
compounding of interest : the return is

Pt2 − Pt1
under discrete compounding
Pt1
or
 Pt2 
ln   under continuous compounding
 Pt 
 1 

• Here we have used discrete time notation

 45
The Log Approximation to Returns
Valid over a short time interval (e.g. 1 day)
• Using the approximation
ln(1+x)  x for small x
we can show that over a short time period t :

Y ( t + δt ) − Y ( t )  Y ( t + δt ) 
 ln  
Y (t )  Y ( t ) 
Similarly
( Pt +δt − Pt )  Pt + δt 
 ln  
Pt  t 
P

Discretely Continuously
compounded return  compounded return
 Using Log Returns 46

• Of course:
 Pt +δt 
ln   = ln ( Pt +δt ) − ln ( Pt )
 Pt 
• Hence it is common practice to approximate discretely
compounded returns over short periods by the continuously
compounded return, i.e. the difference in the log prices
• Daily and intra-day returns are well approximated this way,
but over longer time intervals we often use the exact return.