1. 1.1 Markov Property 1.2 Wiener Process 1.3 2. 2.1 2.2 2.3 2.4 2.5 2.6 Taylor Expansion 2.7 3. Stochastic 3.1 3.2 SDE(Stochastic Differential Equation) 4. Stochastic 4.1 Stochastic integration 4.2 Ito Integral 4.3 Ito Integral 4.4 5. Ito’s Lemma 5.1 Stochastic 5.1.1 5.1.2 5.1.3 First Order Term Second Order Term Cross Product Terms “ ” – Ito Integral Riemann (Ordinary Differential Equation) (Chain rule)
f ( x ) = f ( x0 ) + f x ( x0 )( x − x0 ) + R( x, x0 )
1 1 2 3 f xx ( x0 )( x − x0 ) + f xxx ( x − x0 ) + R( x, x0 ) 2 3! . .
Taylor series expansion
∆x = x − x0 f ( x0 + ∆x ) − f ( x0 ) ≅ f x (∆x ) +
∆x
1 1 2 3 f xx (∆x ) + f xxx (∆x ) 2 3! . x random
random process x deterministic . random
(∆x )2 first-order approximation , ∆x ) “ ” random . 0 , E [∆x ] > 0
2
.
x
(random
(∆x )2
.
.
random process
Taylor series expansion
3.2 SDE
[0, T ]
0 = t0 < t1 < L < tk < Ltn = T k
h
n
.
14
t k − t k −1 = h
t k = kh .
[0, T ]
∆S k = S (kh ) − S ((k − 1)h ) k ∆Wk = [S k − S k −1 ] − E k −1 [S k − S k −1 ] E k −1 [⋅] k −1
S k = S (kh )
.
∆Wk
.
I k −1
(conditional expectation). ∆Wk ∆Wk measurable with respect to I k W0 = 0 . . E k −1 [∆Wk ] = 0 Wk = ∆W1 + L + ∆Wk , Wk martinglale
Ek −1Wk = Ek −1 [∆W1 + L + ∆Wk ]
Ek −1Wk = [∆W1 + L + ∆Wk ] = Wk −1 ∆Wk h .
E [∆Wk ] = σ k2 h
2
, ∆Wk “ ” .
. ,
15
. ∆Wk . E k −1 [S k − S k −1 ] E k −1 [S k − S k −1 ] = A(I k −1 , h ) A(I k −1 , h ) h=0 .
Taylor series expansion
.
A(I k −1 , h ) = A(I k −1 ,0) + a(I k −1 )h + R(I k −1 , h ) h=0 A(I k −1 ,0) = 0 .
h
random
I E k −1 [S k − S k −1 ] ≅ a(I k −1 , kh )h first-order Taylor series approximation stochastic difference equation S kh − S( k −1)h = a(I k −1 , kh )h + σ k Wkh − W(k −1)h . .
[
] h→0 stochastic differential equation . dS (t ) = a(I , t )dt + σ t dW (t )
E [∆Wk ] = σ k2 h
2
∆Wk2 ≅ h
.
W
dW (t )
. W(k −1)h + h − W(k −1)h h
lim h →0
→∞
16
h→0 f (h ) =
.
1 h1 2 = h h dW (t )
Ito Integral
.
17
4. Stochastic
– Ito Integral
. dX t Xt
0
.
= Xt
∫ dX
0
t
u
dX t . stochastic . . t +h
(integral equation)
∫ dX t u
= X t +h − X t
h Ito integral integral t
Wt = ∫ dWu
0 t
dX t random . Wiener process Wt
. stochatic
.
dW
.
(unbounded)
.
4.1 Stochastic integration
Riemann
stochastic difference equation S k − S k −1 = a(S k −1 , k )h + σ (S k −1 , k )∆Wk Riemann .
.
18
∑ [S k − S k −1 ] = ∑ [a(S k −1 , k )h] + ∑ σ (S k −1 , k )[∆Wk ] k =1 k =1 k =1
n −1
n −1
n −1
.
∫
T
0
n ⎧n ⎫ dS u = lim ⎨∑ [a(S k −1 , k )h] + ∑ σ (S k −1 , k )[∆Wk ]⎬, n →∞ k =1 ⎩ k =1 ⎭
T = nh
k −1
random .
(h
).
∫
T
0
a(S u , u )du = lim ∑ [a(S k −1 , k )h] n →∞ k =1
n
Wk − Wk −1
random (convergence)
. .
4.2 Ito Integral
Ito integral 1. σ (S t , t ) 2. σ (S t , t ) “non-explosive” .
. (non-anticipative)
T 2 E ⎡ ∫ σ (S t , t ) dt ⎤ < ∞ . ⎥ ⎢0 ⎦ ⎣
Ito integral .
∫ σ (S , t )dW
T 0 t
t
n→∞
∑ σ (S k =1
n
k −1
, k )[Wk − Wk −1 ] → ∫ σ (S t , t )dWt
T 0
mean square convergence
.
19
Mean Square Convergence X 0 , X 1 ,K, X n ,K random
X
n→∞
. sequence .
Xn
mean square converge
2
lim E [ X n − X ] = 0
Xn − X = εn
(variance)
0
.
∫ σ (S , t )dW
T 0 t
t
Ito integral
2
.
T ⎤ ⎡ n lim E ⎢∑ σ (S k −1 , k )[Wk − Wk −1 ] − ∫ σ (S u , u )dWu ⎥ = 0 0 n→∞ ⎦ ⎣ k =1
5 xt Wiener process .
∫
T
0
xt dt
.
Riemann
V n = ∑ x t i x t i +1 − x t i i =0
n −1
[
] xti x t i +1
non-anticipating
.
Vn Vn = Vn
. 1 ⎡ 2 n −1 2 ⎤ xT − ∑ ∆xti +1 ⎥ 2⎢ i =0 ⎣ ⎦ mean square limit
Z
∑ ∆x i =0
n −1
2 t i +1
mean square limit .
.
⎤ ⎡ n−1 lim E ⎢∑ ∆xt2+1 − Z ⎥ = 0 i n→∞ ⎦ ⎣ i =0
2
Z
.
20
n −1 ⎡ n −1 ⎤ n −1 E ⎢∑ ∆xt2i +1 ⎥ = ∑ E ∆xt2i +1 = ∑ (t i +1 − t i ) = T i =0 ⎣ i =0 ⎦ i =0
[
]
⎤ ⎡ n −1 lim E ⎢∑ ∆xt2i +1 − T ⎥ = 0 n →0 ⎦ ⎣ i =0
2
.1
Vn =
1 ⎡ 2 n −1 2 ⎤ xT − ∑ ∆xti +1 ⎥ 2⎢ i =0 ⎣ ⎦ 1 2 xT − T 2
.
lim E [Vn ] =
2 n→∞
[
]
.
Ito integral
∫
T
0
xt dt =
1 2 xT − T 2
[
]
.
⎤ ⎡ n −1 2 lim E ⎢∑ ∆xti +1 − T ⎥ = 0 n →0 ⎦ ⎣ i =0
2
xt
Wiener process
2
Ito integral
∫ (dx )
T 0 t
Ito integral
T ⎡ n −1 2⎤ lim E ⎢∑ ∆xt2i +1 − ∫ (dxt ) ⎥ = 0 o n →0 ⎦ ⎣ i =0 2
.
T = ∫ dt
0
T
∫ (dx )
T 0 t
2
= ∫ dt
0
T
.
Wt
Wiener process
(dW )2 = dt
.
1
An Introduction to the Mathematics of Financial Derivatives, Salih N. Neftchi pp182-183