0% found this document useful (0 votes)
81 views

Formulae Sheet

This document provides formulae for various statistical distributions and financial concepts. It includes definitions, notations, probability mass and density functions, and key metrics like mean, variance, and standard deviation for distributions like binomial, Poisson, normal and t. It also defines option payoff functions and the binomial option pricing model including parameters like delta, gamma, theta and vega for Greek letters.

Uploaded by

thyanh.vu
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
81 views

Formulae Sheet

This document provides formulae for various statistical distributions and financial concepts. It includes definitions, notations, probability mass and density functions, and key metrics like mean, variance, and standard deviation for distributions like binomial, Poisson, normal and t. It also defines option payoff functions and the binomial option pricing model including parameters like delta, gamma, theta and vega for Greek letters.

Uploaded by

thyanh.vu
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Formulae Sheet

Statistics
sample α-percentile [ n ×α ] th sample of sorted ascending data if n × α is not an integer where
[ ] means “round up to next integer”; average of (n × α )th and
(n × α +1)th samples of sorted ascending data if n × α is an integer
sample interquartile range sample 75% percentile minus sample 25% percentile
n
sample mean absolute deviation
1
n ∑i=1 |x i−sample median|
n
s = n−1 ∑i =1 ( x i − x̄ )
2 1 2
sample variance
sample standard deviation s

sample coefficient of variation s/ x̄


n
sample skewness
1
n−2 ∑i=1 ( xi − x̄ )3 / s 3
n
sample covariance
1
n−1 ∑i=1 ( x i − x̄ )( y i− ȳ )
n
1
n−1 ∑i=1 ( x i− x̄ )( y i − ȳ )

n n
sample correlation
1
n−1 ∑i=1 ( x i− x̄ )2 n−1
1
∑i=1 ( y i− ȳ )2

∑ sign ( ( x i −x j )( y i− y j ) ) /( n2 )
sample Kendall’s tau i< j

Rank ( x i ) Rank ( y i )
sample Spearman’s rho sample correlation between and
Bernoulli Distribution

notation X ~ Ber ( p ) ; 0< p< 1

probability mass function Pr ( X =x )=p x ( 1−p )1− x ; x=0 , 1

mean Ε ( X )=p

variance Var ( X )= p ( 1− p )
Binomial Distribution

notation X ~ Bi ( n , p ) ; n is a positive integer 0< p< 1

probability mass function


()
Pr ( X =x )= n p x ( 1−p )n− x
x ; x=0 , 1 , 2, . .. , n

mean Ε ( X )=n p

variance Var ( X )=n p ( 1− p )


Multinomial Distribution
⃗ ~ M ( n , ⃗p )
X
notation k ; n is a positive integer
k
⃗p= p 1 , p2 ,. .. , pk ; 0< p i <1 ∑i=1 pi=1
n! x x x
Pr ( X 1=x 1 , X 2 =x 2 , .. . , X k =x k ) = p1 1 p2 2 ¿⋅¿ pk k
probability mass function x 1 ! x 2 !⋅¿⋅x k ! ;
k
x i=0 ,1 , 2 ,. .. , n , ∑i=1 x i=n
Ε ( X i )=n× pi
mean
Var ( X i ) =n× p i×( 1−p i )
variance

Poisson Distribution

notation X ~ Poisson ( λ ) ; λ> 0

e−λ λ x
Pr ( X =x )=
probability mass function x! ; x=0 , 1 , 2, . ..

mean Ε ( X )=λ

variance Var ( X )=λ

Poisson Process

notation X ( t ) ~ Poisson ( λ t ) ; λ> 0 t >0

e− λ t ( λ t ) x
Pr ( X ( t )=x )=
probability mass function x! ; x=0 , 1 , 2, . ..

mean Ε ( X ( t ) )=λ t

variance Var ( X ( t ) )= λ t
Discrete Uniform Distribution

notation X ~ U ( m ,n ) ; m≤ n (both are integers)

1
Pr ( X =x )=
probability mass function n−m+1 ; x=m ,m+1, ... , n
1
Ε ( X )= ( m+ n )
mean 2
1
Var ( X )= ( n−m) ( n−m+ 2 )
variance 12

Option Payoff
max ( S t −X , 0 )
exercising a call at time t payoff =
max ( X−S t , 0 )
exercising a put at time t payoff =

Binomial Option Pricing Model


European option on non-dividend-paying share:
e r dt −d
p=
u=e σ √ dt d=e−σ √ dt u−d
where dt is the time period of one-step in the binomial model
f time t ; any node =[ p f time t+1; upper node connected + ( 1− p ) f time t+1; lower node connected ] e−r dt
American option on non-dividend-paying share:
option price at each node = max(backward calculated value, early exercise payoff)

Greek letters:
f 22−f 21 f −f 20
− 21
∂f f 11−f 10 ∂ f f 21−f 00 Γ=
∂Δ

2
S0 u −S 0 S 0− S 0 d
2

Δ= ≈ θ= ≈ ∂S 2
S 0 u + S 0 S 0 + S0 d
2

∂ S S0 u−S0 d ∂t 2 dt 2

2
¿ ¿
∂f f 00 −f 00 ∂ f f 00−f 00
ν= ≈ ρ= ≈
∂σ dσ ∂r dr

Put-call parity on non-dividend-paying share:


c + X e−r T = p+ S 0
Normal Distribution

notation X ~ Ν ( μ , σ2 ); −∞< μ<∞ σ > 0


( x −μ )2

1 2σ 2
f ( x )= e
probability density function σ √2 π ; −∞< x <∞

cumulative distribution function


F ( x )=Φ ( x−μ
σ )

mean Ε ( X )=μ

variance Var ( X )=σ 2

t Distribution

notation
X ~ tn; n is a positive integer
1 Γ( 2 )
n+ 1
1
f ( x )=
√ nπ Γ ( n2 ) 1+ x2
( )
n+1
2

probability density function n ;



−∞< x <∞
Γ ( r ) =∫0 e−s sr−1 ds

mean Ε ( X )=0
n
Var ( X )=
variance n−2 ; n> 2

Exponential Distribution

notation X ~ exp ( λ ) ; λ> 0

probability density function f ( x )= λ e−λ x ; x >0

cumulative distribution function F ( x )=1−e−λ x


1
Ε ( X )=
mean λ
1
Var ( X )=
variance λ2
Gamma Distribution

notation X ~ γ (c , λ ); c >0 , λ> 0

λ c x c−1 e−λ x
f ( x )=
probability density function Γ (c ) ; x >0
c−1
(λ x)j
F ( x )=1−e −λ x
∑ j!
cumulative distribution function j=0 ; if c is an integer
c
Ε ( X )=
mean λ
c
Var ( X )=
variance λ2

Lognormal Distribution

notation X ~ LN ( μ , σ ) ; −∞< μ<∞ σ > 0


(ln x− μ )2

1 2 σ2
f ( x )= e
probability density function xσ √ 2 π ; x >0

cumulative distribution function


F ( x )=Φ ( ln x−μ
σ )
1
μ+ 2 σ 2
mean Ε ( X )=e

Var ( X )=e 2 μ+σ ( e −1 )


2 2
σ
variance

Generalised Extreme Value Distribution

notation X ~ GEV ( μ , σ , ξ ) ; −∞< μ<∞ , σ > 0 , −∞< ξ<∞


1

(
x− μ
)

1

( )
ξ
1 x −μ − −1 − 1+ξ σ
f ( x )= 1+ ξ ξ e
probability density function σ σ
1

( x− μ
)

− 1+ξ ξ
σ
cumulative distribution function F ( x )=e
Gumbel Distribution

notation X ~ GEV ( μ , σ , ξ ) ; −∞< μ<∞ , σ > 0 , ξ=0


x −μ −
x−μ
1 − σ −e σ
f ( x )= e e
probability density function σ ; −∞< x <∞
x−μ

σ
−e
cumulative distribution function F ( x )=e

mean Ε ( X )=μ+0 .5772 σ


σ2 π2
Var ( X )=
variance 6

Fréchet Distribution

notation X ~ GEV ( μ , σ , ξ ) ; −∞< μ<∞ , σ > 0 , ξ >0


1

(
x− μ
)

1

( )
ξ
1 x −μ − −1 − 1+ξ σ σ
f ( x )= 1+ ξ ξ e μ− < x <∞
probability density function σ σ ; ξ
1

( x− μ
)

− 1+ξ ξ
σ
cumulative distribution function F ( x )=e
Γ ( 1−ξ )−1
Ε ( X )=μ+σ
mean ξ ; ξ <1
σ 2 ( Γ ( 1−2 ξ ) −Γ ( 1−ξ )2 ) 1
Var ( X )= 2 ξ<
variance ξ ; 2

Reversed Weibull Distribution

notation X ~ GEV ( μ , σ , ξ ) ; −∞< μ<∞ , σ > 0 , ξ <0


1

(
x− μ
)

1

( )
ξ
1 x −μ − −1 − 1+ξ σ σ
f ( x )= 1+ ξ ξ e −∞< x < μ−
probability density function σ σ ; ξ
1

( x− μ
)

− 1+ξ ξ
σ
cumulative distribution function F ( x )=e
Γ ( 1−ξ )−1
Ε ( X )=μ+σ
mean ξ
σ 2 ( Γ ( 1−2 ξ ) −Γ ( 1−ξ )2 )
Var ( X )=
variance ξ2
Generalised Pareto Distribution
notation X ~ GPD ( μ , σ , ξ ) ; −∞< μ<∞ , σ > 0 , −∞< ξ<∞

( )
1
1 x −μ − −1
f ( x )= 1+ξ ξ
probability density function σ σ ;
x≥μ , ξ=0
x≥μ , ξ >0
σ
μ≤x ≤μ−
ξ, ξ <0

( )
1
x −μ −
F ( x )=1− 1+ξ ξ
cumulative distribution function σ
x −μ
1 − σ
f ( x )= e
probability density function* σ ; ξ=0
x −μ

cumulative distribution function* F ( x )=1−e σ
; ξ=0
σ
Ε ( X )=μ+
mean 1−ξ ; ξ <1
σ2 1
Var ( X )= ξ<
variance (1−ξ )2 ( 1−2 ξ ) ; 2

Multivariate Normal Distribution


⃗ ~ Ν ( ⃗μ , D
X ⃗) ⃗μ=μ 1 , μ 2 ,. . ., μk , −∞< μi < ∞
notation k ;

[ ]
σ 21 σ 1, 2 … σ1 ,k
2
⃗ = σ2 ,1
D σ2 ⋮
⋮ ⋱
σk ,1 ⋯ σ 2k σ i >0 , −∞< σ i, j <∞
;
σ i, j =ρi , j σ i σ j ; −1≤ ρi, j≤1

1 T ⃗ −1 (⃗x−⃗μ )
1 − ( ⃗x −⃗μ ) D
f ( x1 , x 2 , .. . , x k ) = e 2
joint probability density function √|D⃗ |( 2 π ) k
;
−∞< x i <∞
X i ~ Ν ( μ i , σ 2i )
marginal probability distributions
w 1 X 1 + w2 X 2 +. ..+ wn X n ~ Ν ( μoverall , σ 2overall ) n≤k
portfolio return ;
μoverall =w1 μ1 +w 2 μ 2 +. ..+wn μn
σ 2overall=w 21 σ 21 +w 22 σ 22 +. . .+ w2n σ 2n + ∑ w i w j σ i , j
i≠ j
Multivariate t Distribution
⃗ ~ t ( ⃗μ , D
X ⃗) ⃗μ=μ 1 , μ 2 ,. . ., μk , −∞< μi < ∞
notation ν ,k ;

[ ]
σ 21 σ 1, 2 … σ1 ,k
2
⃗ = σ2 ,1
D σ2 ⋮
⋮ ⋱
σk ,1 ⋯ σ 2k σ i >0 , −∞< σ i, j <∞
;

Γ( )
ν +k

( )
ν +k
2 ⃗ −1 ( ⃗x −⃗μ )
( ⃗x −⃗μ )T D −
f ( x1 , x 2 , .. . , x k ) = ν 1+ 2

Γ ( 2 ) √|D
⃗ |( πν )k ν
joint probability density function ;
−∞< x i <∞
X i −μ i
~ tν
marginal probability distributions σi

( )
X i−μi
Ε =0
σi Ε ( X i )=μ i
mean ;

( )
X i−μi ν
Var = ⇒ Var ( X i ) = ν σ 2i
σi ν −2 ν−2 ; ν> 2
variance
ν
Cov ( X i , X j ) = σ
covariance ν−2 i, j ; ν> 2

portfolio return
w 1 X 1 +w2 X 2 +. ..+wn X n for n≤k is not t distributed

Ε ( w1 X 1 +w 2 X 2 +. ..+ wn X n ) =w 1 μ1 + w2 μ2 +. . .+ w n μ n
Var ( w1 X 1 +w 2 X 2 +. . .+ w n X n )

ν
=
ν−2 (
w21 σ 21 +w22 σ 22 +.. .+w 2n σ 2n + ∑ wi w j σ i, j
i≠ j )
Gaussian Copula
F ( x , y )=C ( G ( x ) , H ( y ) )=Φ 2 ( Φ−1 ( G ( x ) ) ,Φ−1 ( H ( y ) ) )

Φ 2 = joint cumulative distribution function of Ν 2 ( ⃗0 , R


⃗)

[ ]
⃗R = 1 r
r 1 ; −1≤r≤1 ,
r =sin
πτ
2
λ U =λ L=0

t Copula
F ( x , y )=C ( G ( x ) , H ( y ) )=t ν , 2 (t −1
ν ( G ( x ) ) ,t ν ( H ( y ) ) )
−1

t ν = cumulative distribution function of t ν

t ν ,2 = joint cumulative distribution function of t ν ,2 ( ⃗0 , ⃗R )

[ ]
⃗R = 1 r
r 1 ; −1≤r≤1 ,
r =sin
πτ
2

λ U =λ L=2−2 t ν +1 (√ ( ν+1 )( 1−r )


1+r )
Clayton Copula
1
−θ
F ( x , y )=C ( G ( x ) , H ( y ) )=ϕ−1 ( ϕ ( G ( x ) ) + ϕ ( H ( y ) ) ) =( ( G ( x ) )−θ + ( H ( y ) )−θ −1 )

t −θ −1 2τ
ϕ ( t )= θ=
θ ; θ> 0 , 1−τ
1
λ U =0 , λ L =2
−θ

Gumbel-Hougaard Copula
1

F ( x , y )=C ( G ( x ) , H ( y ) )=ϕ−1 ( ϕ ( G ( x ) ) + ϕ ( H ( y ) ) ) =e−( ( −lnG ( x )) + (−ln H ( y ))θ )


θ θ

1
θ θ=
ϕ ( t )=(−ln t ) θ≥1 1−τ
1

λ U =2−2 θ , λ L =0
Conditional Probability
Pr ( A∩B )
Pr ( A|B )=
conditional probability Pr ( B )

independence between A and B Pr ( A∩B )=Pr ( A ) Pr ( B ) Pr ( A|B )=Pr ( A )

Bayesian Theory
n
f posterior ( θ|⃗x ) ∝ ∏ f process ( x i|θ ) f prior ( θ )
posterior density function i =1

predictive density function


f predictive ( x¿|⃗x ) =∫ f process ( x¿|θ ) f posterior ( θ| ⃗x ) dθ

Simple Linear Regression


Y i =α + β x i +ε i i=1 , 2 ,. .. , n
ε i ~ Ν ( 0 , σ 2ε )
are independent and identically distributed
Ε ( Y i )=α + β x i Var ( Y i )=σ 2ε

Multivariate Linear Regression


Y i =β 0 +β 1 x 1, i +β 2 x 2, i +. . .+ β k x k , i +ε i i=1 , 2 ,. .. , n
ε i ~ Ν ( 0 , σ 2ε )
are independent and identically distributed
Ε ( Y i )=β 0 + β1 x 1 ,i + β 2 x 2 , i +.. .+ β k x k , i Var ( Y i )=σ 2ε

AR(1) Model
X t =α 0 +α 1 X t −1 + Z t
Z t ’s are independent with mean 0 and variance σ 2

α0 σ2 α 1k σ 2
Ε ( X t )= Var ( X t )= Cov ( X t −k , X t ) =
1−α 1 1−α 21 1−α 21

AR(p) Model
X t =α 0 +α 1 X t −1 + α 2 X t−2 + .. .+α p X t − p +Z t p=1,2,3 ,...
Z t ’s are independent with mean 0 and variance σ 2
α0
Ε ( X t )=
1−α 1 −α 2 −. ..−α p
MA(1) Model
X t =μ +Z t + β Z t −1
Z t ’s are independent with mean 0 and variance σ 2

Ε ( X t )=μ Var ( X t )=( 1+ β2 ) σ 2

Cov ( X t −1 , X t )=β σ 2 Cov ( X t −k , X t ) =0


for k > 1
MA(q) Model
X t =μ +Z t +β 1 Z t−1 + β 2 Z t−2 +.. .+β q Z t−q q=1 ,2,3 ,...
Z t ’s are independent with mean 0 and variance σ 2

Ε ( X t )=μ Var ( X t )=( 1+ β 21 + β 22 +. ..+ β 2q ) σ 2

Forwards
for non-dividend-paying share:

forward price F 0=S 0 e r T


−r ( T −t )
(long) contract value at t ( F t −F 0 ) e
−r ( T −t )
(short) contract value at t ( F 0 −F t ) e

You might also like