1
$\begingroup$

I created paths using GBM to implement The stochastic mesh method. But the method requires the conditional distribution, given some S(t) the probability of S(t+1).

I've searched and can't find this formula, Does someone know it? Is there any other way to calculate the conditional probability?

UPDATE

Let me explain a little better, I'm working on the stochastic mesh method The method require I create N paths over M steps of time. I choose GBM to do this, so in a single path simulation I have

$$ S = S_{0}, S_{1}, ... ,S_{M} $$

Next I need to calculate the continuation value for wich I need Weights

$$ w_{ij} = p_{ij} / \sum_k p_{kj} $$ $$ p_{ij} = P (S_{t + \bigtriangleup t} \in A \mid S{t} = x) $$ The last is the formula I'm looking for the GBM

$\endgroup$
1
$\begingroup$

Applying Itô's lemma to the Black-Scholes SDE and integrating from $t$ to $t+\Delta t$ gives: $$ S_{t+\Delta t} = S_t e^{(r-\frac{1}{2}\sigma^2)\Delta t + \sigma \sqrt{\Delta t}Z} $$ with $Z \sim N(0,1)$, showing that $S_{t+\Delta t}$ given $S_t$ is log-normally distributed.

It is then straightforward to write, for any compact $\mathcal{A} = [a_1,a_2]$ with $0 < a_1 \leq a_2$ and under the risk-netral measure $\mathbb{Q}$

\begin{align} \mathbb{Q}\left(S_{t+\Delta t} \in \mathcal{A} \vert S_t = x\right) &= \mathbb{Q}\left( x e^{ (r-\frac{1}{2}\sigma^2)\Delta t + \sigma \sqrt{\Delta t}Z} \in [a_1, a_2] \right) \\ &= \mathbb{Q} \left( \ln(x) + (r-\frac{1}{2}\sigma^2)\Delta t + \sigma \sqrt{\Delta t}Z \in [\ln(a_1), \ln(a_2)] \right) \\ &= \mathbb{Q} \left( Z \in [a_1^*, a_2^*] \right) \\ &= \Phi(a_2^*) - \Phi(a_1^*) \end{align} where we have used the fact that the natural logarithm is a monotone increasing bijective function and defined $$\Phi(x) = \text{Pr}(X \leq x), X \sim N(0,1)$$ to be the normal cumulative distribution function along with $$ a_i^* = \frac{ \ln\left(\frac{x}{a_i}\right) + \left(r - \frac{1}{2}\sigma^2\right)\Delta t }{\sigma\sqrt{\Delta t}} $$

$\endgroup$
1
$\begingroup$
  1. Note that $${{f}_{W(t)\left| W(s) \right.}}\left(x\left| y \right. \right)=\frac{{{f}_{ W(s),W(t)}}\left( x,y \right)}{{{f}_{ W(s)}}\left( y \right)}=\frac{1}{\sqrt{2\pi(t-s)}}\exp \left[-\frac{{{(x-y)}^{2}}}{2(t-s)} \right]$$

  2. By application of Ito's lemma we have $$ln\,S_{t+\Delta t}=ln\,S_t+\left((\mu-\frac{1}{2}\sigma^2)\Delta t+\sigma(W_{t+\Delta t}-W_t)\Delta t\right)$$ Indeed $ln\,S_{t}$ has a Normal distribution. $$$$

  3. $P(S_{t+\Delta t}<x|S_{t}<y)$=$P(ln\,S_{t+\Delta t}<ln\,x|ln\,S_{t}<ln\,y)$
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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