MATHEMATICAL FINANCE CHEAT SHEET In other words, W (t ) is a drifting Q-Brownian motion with drift (t ) at time t .

Normal Random Variables

Cameron-Martin-Girsanov Converse
A random variable X is Normal N(, 2 ) (aka. Gaussian) under a measure P if and
If (W (t ))t 0 is a P-Brownian motion, and Q is a measure equivalent to P, then there
only if
1 2 2 exists a F -previsible process ((t ))t 0 such that
EP e X = e + 2 , for all real .
 
Z t
A standard normal Z N(0, 1) under a measure P has density Wf(t ) := W (t ) + (s ) d s
Z x 0
1
(x ) = p e x /2 .
2
P[Z x ] = (x ) := (z ) d z . is a Q-Brownian motion. That is, W (t ) plus drift (t ) is a Q-Brownian motion. Ad-
2 ditionally,
Let X = (X 1 , X 2 , . . . , X n )0 with X i N(i , qi i ) and Cov[X i , X j ] = qi j for i , j = 1, . . . , n.
Z t ZT
dQ 1
We call := (1 , . . . , n )0 the mean and Q := (qi j )ni, j =1 the covariance matrix of X . = exp (t ) d W (t ) (t )2 d t .
dP 0
2 0
Assume detQ > 0, then X has a multivariate normal distribution if it has the den-
sity Martingale Representation Theorem
1 1

(x ) = p exp (x )0Q 1 (x ) , x Rn .
Suppose (M (t ))t 0 is a Q-martingale process whose volatility EQ [M (t )2 ] = (t )

(2)n detQ 2
We write X N(,Q ) if this is the case. Alternatively, X N(,Q ) under P if and satisfies (t ) 6= 0 for all t (with Q-probability one). Then if (N (t ))t 0 is any other Q-
RT
only if martingale, there exists an F -previsible process ((t ))t 0 such that 0 (t )2 (t )2 d t <
1

EP [e X ] = exp 0 + 0Q , for all Rn .
0
(with Q-prob. one), and N can be written as
2 Z t
If Z N(0,Q ) and c Rn then X = c 0 Z N(0, c 0Q c ). If C Rmn (i.e., m n matrix) N (t ) = N (0) + (s ) d M (s ),
then X = C Z N(0, C Q C ) and C Q C is a m m covariance matrix.
0 0
0

Gaussian Shifts or in differential form, d N (t ) = (t ) d M (s ). Further, is (essentially) unique.

If Z N(0, 1) under a measure P, h is an integrable function, and c is a constant Multidimensional Diffusions, Quadratic Covariation, and Its Formula
then
EP [e c Z h (Z )] = e c /2 EP [h (Z + c )].
2
If X := (X 1 , X 2 , . . . , X n )0 is a n -dimensional diffusion process with form
Z t Z t
Let X N(0,Q ), h be a integrable function of x Rn , and c Rn . Then
0 1
c 0Q c
X (t ) = X (0) + (s ) d s + (s ) d W (s ),
EP [e c X
h (X )] = e 2 EP [h (X + c )]. 0 0

Correlating Brownian Motions where (t ) Rn m and W is a m -dimensional Brownian motion. The quadration
covariation of the components X i and X j is
Let (W (t ))t 0 and (W
f(t ))t 0 be independent Brownian motions. Given a correla- Z t
tion coefficient [1, 1], define X i , X j (t ) = i (s )0 j (s ) d s ,
c(t ) := W (t ) + 1 2 W
f(t ),
p
W 0
or in differential form d X i , X j (t ) = i (t )0 j (t ) d t , where i (t ) is the i th column
then (W
c(t ))t 0 is a Brownian motion and E[W (t )W
c(t )] = t . Rt
of (t ). The quadratic variation of X i (t ) is X i (t ) = 0 i (s )0 i (s ) d s .
Identifying Martingales The multi-dimensional It formula for Y (t ) = f (t , X 1 (t ), . . . , X n (t )) is
If X t = X (t ) is a diffusion process satisfying f X n
f
d X (t ) = (t , X t ) d t + (t , X t )d W (t ) d Y (t ) = (t , X 1 (t ), . . . , X n (t ))d t + (t , X 1 (t ), . . . , X n (t ))d X i (t )
t i =1
xi
RT
and EP [( 0 (s , X s )2 d s )1/2 ] < (or, (t , x ) c |x | as |x | ), then n
1 X 2f
X is a martingale X is driftless (i.e., (t ) 0 with P-prob. 1). + (t , X 1 (t ), . . . , X n (t ))d X i , X j (t ).
2 i , j =1 xi x j
Novikovs Condition The (vector-valued) multi-dimensional It formula for
In the case d X (t ) = (t )X (t ) d W (t ) for some F -previsible process ((t ))t 0 , then Y (t ) = f (t , X (t )) = (f1 (t , X (t )), . . . , fn (t , X (t )))0
we have the simpler condition
ZT where fk (t , X ) = fk (t , X 1 , . . . , X n ) and Y (t ) = (Y1 (t ), Y2 (t ), . . . , Yn (t ))0 is given component-
1 wise (for k = 1, . . . , n ) as
EP exp (s )2 d s < X is a martingale. n
2 0 fk (t , X (t )) X fk (t , X (t ))
d Yk (t ) = dt + d X i (t )
Its Formula t i =1
xi
n
For X t = X (t ) given by d X (t ) = (t ) d t + (t ) d W (t ) and a function g (t , x ) that is 1 X 2 fk (t , X (t ))
+ d X i , X j (t ).
twice differentiable in x and once in t . Then for Y (t ) = g (t , X t ), we have 2 i , j =1 xi x j
g g 1 2g
d Y (t ) = (t , X t ) d t + (t , X t ) d X t + (t )2 (t , X t )d t . Stochastic Exponential
t x 2 x2
The Product Rule The stochastic exponential of X is Et (X ) = exp(X (t ) 21 X (t )). It satisfies
Given X (t ) and Y (t ) adapted to the same Brownian motion (W (t ))t 0 , E (0) = 1, E (X )E (Y ) = E (X + Y )e X ,Y , E (X )1 = E (X )e X ,X .
d X (t ) = (t )d t + (t )d W (t ), d Y (t ) = (t ) d t + (t ) d W (t ). The process Z = E (X ) is a positive process and solves the SDE
Then d (X (t )Y (t )) = X (t ) d Y (t ) + Y (t ) d X (t ) + d X , Y (t ). dZ =Z dX , Z (0) = e X (0) .
| {z }
(t )(t ) d t
Solving Linear ODEs
In the other case, if X (t ) and Y (t ) are adapted to two different and independent
Brownian motions (W (t ))t 0 and (W f(t ))t 0 , The linear ordinary differential equation
d X (t ) = (t ) d t + (t ) d W (t ), d Y (t ) = (t ) d t + (t ) d W f(t ). d z (t )
= m (t ) + (t )z (t ), z (a ) = ,
Then d (X (t )Y (t )) = X (t ) d Y (t ) + Y (t ) d X (t ) as d X , Y (t ) = 0. dt
for a t b has solution given by
Radon-Nikodm Derivative Z t Z t 
Given P and Q equivalent measures and a time horizon T , we can define a random z (t ) = t + t 1
u m (u) d u , t := exp (u ) d u ,
variable dd QP defined on P-possible paths, taking positive real values, such that a a

Z t
Z t
Z t
dQ

EQ [X T ] = EP X T , for all claims X T knowable by time T , = exp (u) d u + m (u ) exp (r ) d r d u .
dP a a u
EQ [X t |Fs ] = 1
s EP
[t X t |Fs ], for s t T ,
Solving Linear SDEs
where t is the process EP [ dd QP |Ft ].
The linear stochastic differential equation
Cameron-Martin-Girsanov Theorem d Z (t ) = [m (t ) + (t )Z (t )]d t + [q (t ) + (t )Z (t )] d W (t ), Z (a ) = ,
If (W (t ))t 0 is a P-Brownian motion and ((t
R))Tt 0 is an F-previsible process satis- for a t b has solution given by
fying the boundedness condition EP exp 12 0 (t )2 d t < , then there exists a
Z t Z t

measure Q such that: Z (t ) = Et + Et Eu1 [m (u) q (u)(u)] d u + Et Eu1 q (u) d W (u ),

a a
Q is equivalent
Z to P, Rt Rt
T ZT where Et := Et (X ) and X (t ) = (u ) d u + (u )d W (u ). In other words,
dQ 1 a a
= exp (t ) d W (t ) (t )2 d t , Z t Z t Z t 
dP 2 0 1
0 Et = exp (u) d u + (u) d W (u) (u)2 d u .
f(t ) := W (t ) + t (s ) d s is a Q-Brownian motion. 2 a
R
W 0
a a
 Fundamental Theorem of Asset Pricing AJD option pricing

Let X be some FT -measurable claim, payable at time T . The arbitrage-free price Define the Fourier transform inversion of the conditional expectation
V of X at time t is
ZT
1b X
T
ZT G (a , b , y ) = EQ exp R (X s )d s e a XT

 T y
V (t ) = EQ exp r (s ) d s X Ft , 0

(a , X 0 , T ) 1 ((a + i v b , X 0 , T )e i v y )
t
Z
where Q is the risk-neutral measure. = dv
2 0
v

Market Price Of Risk The i th entry in X is the log asset price and k = l o g (K ), the log strike. d is a vector
whose i th element is 1, else zero. The corresponding call option price is
Let X t = X (t ) be the price of a non-tradable asset with dynamics d X (t ) = (t ) d t + C = G (d , d , k ) K G (0, d , k )
(t )d W (t ) where ((t ))t 0 and ((t ))t 0 are previsible processes and (W (t ))t 0 is
a P-Brownian motion. Let Y (t ) := f (X t ) be the price of a tradable asset where The Heath-Jarrow-Morton Framework
f : R R is a deterministic function. Then the market price of risk is Given a initial forward curve T 7 f (0, T ) then, for every maturity T and under the
t f 0 (X t ) + 12 2t f 00 (X t ) r f (X t ) real-world probability measure P, the forward rate process t 7 f (t , T ) follows
(t ) := , Z t Z t
t f 0 (X t )
f (t , T ) = f (0, T ) + (s , T ) d s + (s , T )0 d W (s ), t T,
and the behaviour of X t under the risk-neutral measure Q is given by 0 0
r f (X t ) 12 2t f 00 (X t ) where (t , T ) R and (t , T ) := (1 (t , T ), . . . , n (t , T )) satisfy the technical condi-
d X (t ) = (t )d W
f(t ) + dt. RT RT
f 0 (X
t) tions: (1) and are previsible and adapted to Ft ; (2) 0 0 |(s , t )| d s d t <
for all T ; (3) sups ,t T k(s , t )k < for all T . The short-rate process is given by
Blacks Model Z t Z t
r (t ) = f (t , t ) = f (0, t ) + (s , t ) d s + (s , t ) d W (s ),
Consider a European option with strike price K on a asset with value VT at ma- 0 0
turity time T . Let FT be the forward price of VT , F0 the current forward price. If
so the cash account and zero coupon T -bond prices are well-defined and obtained
log VT N(F0 , 2 T ) then the Call and Put prices are given by through
C = P (0, T )(F0 (d 1 ) K (d 2 )), P = P (0, T )(K (d 2 ) F0 (d 1 )), Z t  ZT
log(EQ (VT )/K ) + 2 T /2 p B (t ) = exp r (s ) d s , P (t , T ) = exp f (t , u) d u .
where d 1 = p and d 2 = d 1 T . 0 t
T
The discounted asset price Z (t , T ) = P (t , T )/B (t ) satisfies
ZT
Forward Rates, Short Rates, Yields, and Bond Prices 1
d Z (t , T ) = Z (t , T ) S 2 (t , T ) (t , u) d u d t + S (t , T )0 d W (t ) ,
The forward rate at time t that applies between times T and S is defined as 2 t
| {z }
1 P (t , T ) b (t ,T )
F (t , T,S ) = log .
S T P (t ,S ) where S (s , T ) :=
RT
(s , u) d u . The HJM drift condition states that
s
The instantaneous forward rate at time t is f (t , T ) = limS T F (t , T,S ). The instan-
taneous risk-free rate or short rate is r (t ) = limT t f (t , T ). The cash account is Q is EMM (i.e., no arbitrage for bonds) b (t , T ) = S (t , T )(t )0 ,
given by f(t ) := W (t ) t (s ) d s is a Q-Brownian motion. If this holds, then under
R
Z t  where W 0
B (t ) = exp r (s ) d s , Q, the forward rate process follows
Z t ZT Z t
0
and satisfies d B (t ) = r (t )B (t ) d t with B (0) = 1. The instantaneous forward rates f (t , T ) = f (0, T ) + (s , T ) (s , u)0 d u d s + (s , T ) d W
f(s ),
and the yield can be written in terms of the bond prices as 0
|
s
{z }
0

log P (t , T ) HJM drift

f (t , T ) = log P (t , T ), R (t , T ) = .
T T t and the discounted asset Z (t , T ) satisfies d Z (t , T ) = Z (0, T )Et (X ) with
Conversely, Z t
ZT X (t ) = S (s , T )0 d W
f(s ).
P (t , T ) = exp f (t , u) d u and P (t , T ) = exp((T t )R (t , T )). 0
t The LIBOR Market Model
For a tenor > 0, the LIBOR rate L (T, T, T + ) is the rate such that an investment
Affine Jump Diffusion (AJD) Models
of 1 at time T will grow to 1+L (T, T, T +) at time T +. The forward LIBOR rate
The state vector X t follows a Markov process solving the SDE (i.e., a contract made at time t under which we pay 1 at time T and receive back
1 + L (t , T, T + ) at time T + ) is defined as
d X t = (X t )d t + (X t )d Wt + d Z t
1 P (t , T )

where W is an adapted Brownian, and Z is a pure jump process with intensity . L (t , T ) := L (t , T, T + ) = 1 ,
The moment generating function of the jump sizes is (c ) = EQ (exp(c J )). Impose P (t , T + )
an affine structure on , T , and the discount rate R , possibly time dependent: and satisfies L (T, T ) = L (T, T, T + ).
Under the real-world probability measure P, The LMM assumes that each LIBOR
(x ) = K 0 +K 1 x ((x )(x )T )i j = (H0 )i j +(H1 )i j x (x ) = L 0 +L 1 x R (x ) = R0 +R1 x process (L (t , Tm ))0t T satisfies
m
Given X 0 , the risk neutral coefficients (K , H , L , , R ) completely determine the dis- d L (t , Tm ) = L (t , Tm ) (t , L (t , Tm )) d t + m (t , L (t , Tm ))0 d W (t ) ,
 
counted risk neutral distribution of X . Introduce the transform function
where W = (W 1 , . . . , W d ) is a d -dimensional Brownian motion with instantaneous
ZT 

R (X s )d s e u X T  F0 = e (0,u)+ (0,u) x0
T T
(u, X 0 , T ) = EQ exp
 correlations
0
 d W i , W j (t ) = i , j (t ) d t , i , j = 1, 2, . . . , d .
where and solve the Ricatti ODEs subject to (T, u) = 0, (T, u) = u: The function (t , L ) : [0, Tj ] R RN d is the volatility, and (t ) : [0, Tj ] R is the
drift.
(t , u) =K 1T (t , u) + 21 (t , u)T H1 (t , u) + L 1 ( ( (t , u)) 1) R1 Let 0 m , n N 1. Then the dynamics of L (t , Tm ) under the forward measure
(t , u) =K 0T (t , u) + 21 (t , u)T H0 (t , u) + L 0 ( ( (t , u)) 1) R0 PTn+1 is for m < n given by
n

X
AJD bond pricing d L (t , Tm ) = L (t , Tm ) (t , Tm ) Tr ,Tr +1 (t )0 d t + (t , Tm )d W m (t )
r =m +1
In , set L i = R0 = u = 0, R1 = 1 to obtain the zero coupon bond with maturity For m = n,
T t via the Ricatti ODEs: d L (t , Tm ) = L (t , Tm )(t , Tm )d Wt m
and for m > n we have
Short rate model K0 K1 H0 H1 P?MR? m

X
Merton 2 NN d L (t , Tm ) = L (t , Tm ) (t , Tm ) Tr ,Tr +1 (t )0 d t + (t , Tm )d Wt m
Dothan 2 YN r =n+1
Vasicek 2 NY
CIR 2 YY
Pearson-Sun 2 2 YY
Ho & Lee (t ) 2 NN
Hull & White (t ) 2 NY
Extended Vasicek (t )(t ) (t ) (t )2 NY
Black-Karasinski (t )(t ) (t ) (t )2 YY

P means the process stays positive, MR means rt is mean-reverting. Closed form so-
lutions for bond prices and European options exist for all models except for , which
github.com/daleroberts/math-finance-cheat-sheet
describes the evolution of d log(rt ) instead of d rt .