Explicit formula for product rule for Ito processes

600 Views Asked by At

We know that we have the following formula $d(XY)=XdY+YdX+d[X,Y]$, but how does this look when we have two actual processes?

Should it look something like this?

$d(XY)=XdY+YdX+d[X,Y]=X(\mu_{Y}dt+\sigma_{Y}dW)+Y(\mu_{X}dt+\sigma_{X}dW)+\sigma_{X}\sigma_{Y}dt=X \mu_{Y}+Y \mu_{X}+\sigma_{X}\sigma_{Y}dt +(\sigma_{Y}+\sigma_{X})dW=(\mu_{X}dt+\sigma_{X}dW )\mu_{Y}+(\mu_{Y}dt+\sigma_{Y}dW) \mu_{X}+\sigma_{X}\sigma_{Y}dt +(\sigma_{Y}+\sigma_{X})dW=$

$(\mu_{X}+\mu_{Y}+\sigma_{X}\sigma_{Y})dt+(\sigma_{X}\mu_{Y}+\sigma_{Y}\mu_{X}+\sigma_{Y}+\sigma_{X})dW$

where I used the definition of integration with respect to a semimartingale and the formula for the crossvariation of two Ito processes. Is the above equalites true for some hypothesis on the processes?

1

There are 1 best solutions below

1
On BEST ANSWER

To be more general, let us suppose that we have two Ito process $X$ and $Y$ (respecting the usual conditions). Their dynamics is defined as follows: \begin{align} dX_t &= \mu_X(t,X_t)dt + \sigma_X(t,X_t)dW_t^X \\ dY_t &= \mu_Y(t,Y_t)dt + \sigma_Y(t,Y_t)dW_t^Y \end{align} where $(W^X,W^Y)$ are standard Brownians with correlation $d\langle{W_.^X,W_.^Y}\rangle_t=\rho_{XY}dt$. Applying the integration by parts, we have: \begin{align} d(X_tY_t) &= Y_tdX_t + X_tdY_t + d\langle{X_.,Y_.}\rangle_t \\ &= Y_t\left[\mu_X(t,X_t)dt + \sigma_X(t,X_t)dW_t^X\right] + X_t\left[\mu_Y(t,Y_t)dt + \sigma_Y(t,Y_t)dW_t^Y\right] + \sigma_X(t,X_t)\sigma_Y(t,Y_t)d\langle{W_.^X,W_.^Y}\rangle_t \\ &= Y_t\left[\mu_X(t,X_t)dt + \sigma_X(t,X_t)dW_t^X\right] + X_t\left[\mu_Y(t,Y_t)dt + \sigma_Y(t,Y_t)dW_t^Y\right] + \\&\sigma_X(t,X_t)\sigma_Y(t,Y_t)\rho_{XY}dt\\ \end{align} As far I know, that's all we can say.