Let $(\Omega,F,P)$ be a probability space with filtration $\left\{F_{t}\right\}_{t\geq 0}$ generated by one dimensional Brownian motion $(B_{t})_{t\geq 0}$ defined on $(\Omega,F,P)$, assume that $\theta$ is a real constant and

$$Z_t := \exp \left(\theta B_t - \frac{1}{2}\theta^2 t\right)$$

Check the martingale property with respect to filtration $\left\{F_{t}\right\}_{t\geq 0}$ for the following processes:

I. $(B_{t})_{t\geq 0}$

II. $(Z_{t})_{t\geq 0}$

My attempt:

I. We have to show: $E[B_t | \mathcal{F}_s] = B_s$, so

$E[B_t | \mathcal{F}_s] = E[B_t - B_s + B_s | \mathcal{F}_s] = B_s + E[B_t - B_s | \mathcal{F}_s] = B_s + 0 = B_s$

My question: Why is $E[B_t - B_s | \mathcal{F}_s] = 0$?

II. We have to show $E[Z_t | \mathcal{F}_s] = Z_s$, so

$$ E[Z_t | \mathcal{F}_s] = E[e^{\theta (B_t - B_s) + \theta B_s}e^{e^-\frac{1}{2}\theta ^2(t-s) - \frac{1}{2} \theta^2s}| \mathcal{F}_s]== E[Z_s e^{\theta (B_t - B_s)}| \mathcal{F}_s]e^{-\frac{1}{2}\theta^2 (t-s)} $$

$$= Z_s E[e^{\theta (B_t - B_s)}| \mathcal{F}_s]e^{-\frac{1}{2}(t-s)\theta^2}= Z_s e^{\frac{1}{2}(t-s) \theta^2}e^{-\frac{1}{2}(t-s) \theta^2}=Z_s$$

My question: Why is $E[e^{\theta (B_t - B_s)}| \mathcal{F}_s] = e^{\frac{1}{2}(t-s) \theta^2}$ ?

  • $\begingroup$ $B_t-B_s|\mathcal{F}_s$ has normal distribution with mean $0$ and variance $t-s$ $\endgroup$ – gt6989b Mar 14 '16 at 14:32
  • $\begingroup$ But then why is $E[e^{B_t}] = e^{\frac{t}{2}}$ and why is the $\theta$ squared? Thanks btw! $\endgroup$ – Ptru Mar 14 '16 at 15:14
  • $\begingroup$ $e^{B_t}$ has lognormal distribution $\endgroup$ – gt6989b Mar 14 '16 at 16:16
  • $\begingroup$ You can calculate $E[e^{\theta B_t}]$ via Ito's formula. $\endgroup$ – Xuxu Oct 6 '16 at 0:23

Assume that $F_{s}=\sigma\left(B_{i}\ \colon\ 0\leq i\leq s\right)$.

Then for every $0\leq s\leq t$ random variable $B_{t}-B_{s}$ is independent of $F_{s}$ and random variable $B_{s}$ is $F_{s}$ - measurable.

We also know that: $B_{u}\sim \mathcal{N}(0,u)$

Check basic properties of conditional expectation: link

I. $$\mathbb{E}\left[B_{t}|F_{s}\right]=\mathbb{E}\left[B_{t}-B_{s}+B_{s}|F_{s}\right]=\mathbb{E}\left[B_{t}-B_{s}|F_{s}\right]+\mathbb{E}\left[B_{s}|F_{s}\right]$$



$$\mathbb{E}\left[Z_{t}|F_{s}\right]=\mathbb{E}\left[e^{\theta B_{t}-\frac{1}{2}\theta^{2}t}|F_{s}\right]=e^{-\frac{1}{2}\theta^{2}t}\mathbb{E}\left[e^{\theta B_{t}}|F_{s}\right]=e^{-\frac{1}{2}\theta^{2}t}\mathbb{E}\left[e^{\theta (B_{t}-B_{s})+\theta B_{s}}|F_{s}\right]$$

$$=e^{\theta B_{s}-\frac{1}{2}\theta^{2}t}\mathbb{E}\left[e^{\theta (B_{t}-B_{s})}|F_{s}\right]=e^{\theta B_{s}-\frac{1}{2}\theta^{2}t}\mathbb{E}\left[e^{\theta (B_{t}-B_{s})}\right]$$

$$=e^{\theta B_{s}-\frac{1}{2}\theta^{2}t}\cdot e^{\frac{1}{2}\theta^{2}(t-s)}=e^{\theta B_{s}-\frac{1}{2}\theta^{2}s}=Z_{s}$$




$$\mathbb{E}\left[e^{\theta (B_{t}-B_{s})}\right]=\int_{\Omega}e^{\theta (B_{t}-B_{s})}d\mathbb{P}=\int_{\mathbb{R}}e^{\theta x}\frac{1}{\sqrt{2\pi (t-s)}}e^{-\frac{x^{2}}{2(t-s)}}dx$$

$$=\int_{\mathbb{R}}\frac{1}{\sqrt{2\pi (t-s)}}exp\left(\frac{-\left(x-\theta(t-s)\right)^{2}+\theta^{2}(t-s)^{2}}{2(t-s)}\right)dx$$ $$=e^{\frac{1}{2}\theta^{2}(t-s)}\int_{\mathbb{R}}\frac{1}{\sqrt{2\pi (t-s)}}exp\left(\frac{-\left(x-\theta(t-s)\right)^{2}}{2(t-s)}\right)dx$$

$$=e^{\frac{1}{2}\theta^{2}(t-s)}\cdot 1= e^{\frac{1}{2}\theta^{2}(t-s)}$$


$$f(x)=\frac{1}{\sqrt{2\pi (t-s)}}exp\left(\frac{-\left(x-\theta(t-s)\right)^{2}}{2(t-s)}\right)$$

is probability density function of random variable with normal distribution of the form: $$\mathcal{N}\left(\theta(t-s),t-s\right)$$


$$\int_{\mathbb{R}}f(x)\ dx=1$$


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.