Best predictor of Brownian motion

37 Views Asked by At

Let $B_t$ be brownian motion at time $ t \geq 0$.

Then I want to find the best predictor of $B_8 + 4$ given that there are observations of brownian motion up to time $t = 1$.

Approach:

Essentially, if $X = X_8 +4$, then $Y$ is our observation, we want to find $\hat X = \mathbb{E}(X|Y)$

So:

$$\mathbb{E}(B_8 + 4 | B_1) = \mathbb{E}(B_8|B_1) + 4$$ since $4$ is a constant.

$$\mathbb{E}(B_8 - B_1 +B_1 | B_1) + 4 = \mathbb{E}(B_8 - B_1|B_1) + \mathbb{E}(B_1|B_1)$$

$$\mathbb{E}(B_8-B_1|B_1) = 0$$ since $B_8-B_1$ is independent of $B_1$.

and $\mathbb{E}(B_1|B_1) = B_1$

Therefore the best predictor is $B_1 + 4$.

Is this correct?