Right now I dont know how to find the expectation and the variance for the following stochastic integral: $$\int_{0}^t B(s) dW(s)$$ where $B(t)$ and $W(t)$ are correlated standard Brownian Motions with $d\langle B, W \rangle_t = \gamma dt$.

Usually, when $B(t)$ and $W(t)$ are independent, we know $$\mathbb{E}\left(\int_{0}^t B(s) dW(s)\right) = 0$$ and by Ito's isometry, $$Var\left(\int_{0}^t B(s) dW(s)\right) = \mathbb{E}\left(\int_{0}^t B(s) dW(s)\right)^2 = \int_{0}^t \mathbb{E}B(s)^2 ds.$$

Does these results still apply?


Decompose your Brownian motion $W$ as $W_t = \gamma B_t + \bar \gamma B_t^\bot$, where $\bar \gamma = \sqrt{1-\gamma^2}$ and $B^\bot$ is independent from $B$. Then:

$$\int_0^t B_s \, \mathrm dW_s = \gamma \int_0^t B_s \, \mathrm dB_s + \bar\gamma \int_0^t B_s \, \mathrm dB_s^\bot$$

The first term can be computed using Itô's formula, the second one involves two independent Brownian motions.


One other way to understand the integral $\int_0^tB(s)\,dW(s)$ is to understand the $(B(t),W(t))^{\top}$ as a 2-d Wiener process with covariance function $\begin{pmatrix}t&\gamma t\\ \gamma t& t\end{pmatrix}$. Using these notations, the Ito's isometry is effective too and following equalities are held: $$ \mathsf{E}\Bigl[\int_0^tB(s)\,dW(s)\Bigr]=0,\qquad \mathsf{E}\Bigl(\int_0^tB(s)\,dW(s)\Bigr)=\int_0^t\mathsf{E}[B(s)]^2\,ds=\frac{t^2}2.$$

  • $\begingroup$ pretty old question, but still, isn't there a missing square in the last line? The stochastic integral should be squared inside the expectation sign. $\endgroup$ – RScrlli Jan 30 at 8:16

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.