Scribd Logo
Upload

Welcome to Scribd!

  • 9 views

    Presentation Asian Options

    Uploaded by

    edem
    This document summarizes three Monte Carlo schemes for pricing Asian options: 1. The standard scheme discretizes the asset path but has low accuracy. 2. The first higher accuracy scheme uses the trapezoidal rule and conditional expectations, improving over the standard scheme. 3. The second higher accuracy scheme further uses Taylor expansions to integrate the asset path, achieving even higher accuracy. Numerical results show the standard scheme is less efficient than the two improved schemes, which provide increasingly accurate option prices using more time steps.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Presentation Asian Options

    Uploaded by

    edem
    9 views8 pages
    This document summarizes three Monte Carlo schemes for pricing Asian options: 1. The standard scheme discretizes the asset path but has low accuracy. 2. The first higher accuracy scheme uses the trapezoidal rule and conditional expectations, improving over the standard scheme. 3. The second higher accuracy scheme further uses Taylor expansions to integrate the asset path, achieving even higher accuracy. Numerical results show the standard scheme is less efficient than the two improved schemes, which provide increasingly accurate option prices using more time steps.

    Original Description:

    Document portant sur des recherches sur les options asiatiques

    Original Title

    Presentation_Asian_options

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes three Monte Carlo schemes for pricing Asian options: 1. The standard scheme discretizes the asset path but has low accuracy. 2. The first higher accuracy scheme uses the trapezoidal rule and conditional expectations, improving over the standard scheme. 3. The second higher accuracy scheme further uses Taylor expansions to integrate the asset path, achieving even higher accuracy. Numerical results show the standard scheme is less efficient than the two improved schemes, which provide increasingly accurate option prices using more time steps.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    9 views8 pages

    Presentation Asian Options

    Uploaded by

    edem
    This document summarizes three Monte Carlo schemes for pricing Asian options: 1. The standard scheme discretizes the asset path but has low accuracy. 2. The first higher accuracy scheme uses the trapezoidal rule and conditional expectations, improving over the standard scheme. 3. The second higher accuracy scheme further uses Taylor expansions to integrate the asset path, achieving even higher accuracy. Numerical results show the standard scheme is less efficient than the two improved schemes, which provide increasingly accurate option prices using more time steps.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 8

    Monte Carlo methods for the pricing of Asian options

    The price of an Asian option with maturity T can be written:


    r (T t)
    P(t, S, A) = e Ef (St , AS (0, t))

    where Z t
    1
    AS (0, t) = Su du
    t 0

    I No closed-form analytical solution for the arithmetic Asian


    options.
    For an arithmetic Asian call option with fixed strike:

    f (s, a) = (a K )+
    1. Standard scheme

    The step size: h = T T


    N , the times tk = k N = kh,
    The number of simulation of Monte Carlo: M
    n 1
    X
    YTr ,n = h Stk
    k=0

    The price at maturity of a fixed strike Asian call option is


    M
    X n 1
    e rT 1 X
    P(t, S, A) = (Stk K )+
    M N
    j=1 k=0
    2. First Higher accuracy scheme

    For tk  u  tk+1 :

    L Wu | Wtk = x, Wtk+1 = y
    ✓ ◆
    tk+1 u u tk (tk+1 u) (u tk )
    =N x+ y,
    h h h
    Using this law, let h = tk+1 tk , we can compute
    Z T Z T
    1 1
    E( Su du|Bh ) = E(Su |Bh )du
    T 0 T 0

    n 1Z ⇣ ⌘ ⇣ ⌘2
    1 X tk+1 u tk
    (Wtk+1 Wtk )
    2 u tk
    +ru Wtk
    2
    t
    = ... = e h 2 h
    e 2 k du
    T tk
    k=0
    2. First Higher accuracy scheme

    Using Taylor expansion to simplify it, we have


    n 1
    1 X rh Wtk+1 Wtk
    YTr ,n = Stk (1 + + )
    N 2 2
    k=0

    The approximate price is given by:


    M
    X n 1
    e rT 1 X rh Wtk+1 Wtk
    P(t, S, A) = (Stk (1 + + ) K )+
    M N 2 2
    j=1 k=0
    2. First Higher accuracy scheme

    This scheme is equivalent to the trapezoidal method,


    n 1
    1 X
    YTt,n = (Stk + Stk+1 )/2
    N
    k=0

    Proof:
    n 1 ✓ ◆
    1 Xh 2
    YTt,n = St e (Wtk+1 Wtk ) 2
    h+rh
    +1
    T 2 k
    k=0

    n 1
    1 Xh
    = St (1 + 1 + Wtk+1 Wtk + rh + O(h(Wtk+1 Wtk )))
    T 2 k
    k=0

    = YTr ,n
    3. Second Higher accuracy scheme

    In a similar way, using Taylor expansion, we have


    n 1 Z tk+1
    1 X rh2
    YTp,n = Stk (h + + (Wu Wtk )du)
    T 2 tk
    k=0

    Knowing the conditionalR t law of Wu , we can compute the explicit


    formula of the law of ( tkk+1 Wu du|Wtk = x, Wtk+1 = y ).
    Z tk+1
    x + y h3
    L( Wu du|Wtk = x, Wtk+1 = y ) = N ( h, )
    tk 2 6
    Variance Reduction - Control Variate

    RT RT
    We approximate T1 0 St dt by ZT = exp( T1 0 ln(St )dt).
    RT
    ln(ZT ) = T1 0 ln(St )dt has a normal law:
    2T 2T
    L(ln(ZT )) = N ((r , ) )
    2 2 3
    We use ✓ ⇣h 2
    i RT ⌘ ◆
    T
    rT r +T Wt dt
    e S0 e 2 2 0
    K
    +
    as a control variable.
    Note that each control variable has to be adapted to the schemes.
    Numerical computation

    We test the three schemes with the following values:


    S0 = 100, K = 100, T = 1, r = 0.1, = 0.2, m = 100000.

    Scheme Time steps Price Confidence intervals at 95%


    10 5.943 [5.943, 5.952]
    (1) 50 6.843 [6.835, 6.852]
    100 6.961 [6.953, 6.969]
    10 6.570 [6.562, 6.579]
    (2) 50 6.898 [6.890, 6.907]
    100 6.967 [6.959, 6.976]
    10 6.513 [6.505, 6.522]
    (3) 50 6.955 [6.947, 6.964]
    100 6.970 [6.962, 6.978]

    I Standard scheme is not very efficient compared to the others

    You might also like