MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I need the martingale part of the put payoff (not $C^2$..). Where $S_t=exp(\sigma W_t -\frac{\sigma^2t}{2})$

$d[(S_t -K)^+ ]$ ??

I guess I need to use local times but how?

share|cite|improve this question
up vote 2 down vote accepted

Thanks you all!

(proof, for $\phi(t,S_t)=(K-S_t)^+$:

Step 1 smoothing : $\phi_\epsilon(x)=1_{S_t\leq K+\epsilon}\cdot\phi(x)+1_{S_t\in]K-\epsilon,K]} \cdot \psi(x)$, where $\psi(x)=-\frac{1}{\epsilon^2}(K-x)^2(K-x-2\epsilon)$ .

This function is $C^1$, and also $C^2$ excepting in a countable set.

Step 2 Itô on $\phi_\epsilon(S_t)= \phi_\epsilon(S_0)+\int^t_0\phi_\epsilon^{'}(S_t)dS_t+\frac{1}{2}\int^t_0 1_{S_t\in[K-\epsilon,K]}\phi_\epsilon^{''}(S_t)d\langle S\rangle _t$ because $\phi _\epsilon=0$ out of $[K-\epsilon,K]$

Let's denote by $L_t=lim_{\epsilon \to0}{\frac{1}{2\epsilon^2}*\int{_{K-\epsilon}^K(3S_t+4\epsilon-3K)dS_t}}$ it's a finite variation process since it is increasing

Step 3 We have that $\phi_\epsilon(S_t)-\phi_\epsilon(S_0)-\int^t_0\phi_\epsilon^{'}(S_t)dS_t \space \xrightarrow {L^2} \space\phi(S_t) -\phi(S_0) -\int^t_0\phi^{'}(S_t)dS_t $ (because $\int^t_0\phi_\epsilon^{'}(S_t) 1_{S_u\in[0,K-\epsilon]}dS_u \xrightarrow {L^2}\int_0^t\phi^{'}(S_u)du$ by using Itô isometry)

Finally We get the formula 'à la bridge' namely

$(K-S_t)^+=(K-S_0)^+-\int_0^t1_{K\leq S_u}dS_u+L_t$

and the martingale part is

$(K-S_0)^+-\int_0^t1_{K\leq S_u}\sigma S_u dB_u$

share|cite|improve this answer
I think that the half factor difference on the Local Ltime term that we have is about the way Local Time is deifned. Different "autors" use different conventions (multiplying $L_t$ by a factor 2 or not in their definition). The definition I use in my answer doesn't use the factor "2" and yours use it. Regards. – The Bridge Mar 7 '11 at 17:07


Simply use Itô-Tanaka's formula I guess this should give something like : $df(S_t)=D_-f(S_t)dS_t+\frac{1}{2}dL^s_tf''(ds)$

with $f(S)=(S-K)^+$ so $D_-f(S)=1_{]K,+\infty}(S)$ and $f''(ds)=\delta_K(ds)$

This gives if I am not mistaken :


With $L^K_t$ being the local time of your geometric Brownian Motion $S$ around level $K$ at time $t$.


Edit NB:
-$D_-$ stands for the left derivative of $f$

-$f''(ds)$ stands for second derivatives in the distribution-sense.

-The use of Itô-Tanaka's formula allows to avoid the derivation of the Mollifiers-type argument for the direct proof of the result (which is quite cumbersome in my opinion). I should add that Ito-Tanaka's formula is applicable to every $f$ that is the differnce of two convex functions if I remember well, which is the case here with $f(x)=(X-K)^+$.

share|cite|improve this answer
So, the martingale part is $\sigma\int1_{[K,\infty)}(S)S\,dW$, which was the question. – George Lowther Mar 6 '11 at 17:02
@george lowther: that's right thanks for the precision – The Bridge Mar 6 '11 at 18:50

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.