I would be grateful for some help with the following exercise:

Let $(X_n ,n≥1)$ be a sequence of independent random variables with $E[X_i]=0$, and $Var(X_i)=σ_i^2<\infty, ∀i ∈\mathbb{N}$. Prove that $S_n^2 − Var(S_n)$ is a martingale, where $S_n:= \sum\limits_{i=1}^n X_i$

We have to show: $E[S_{n+1}^2-Var(S_{n+1})|\mathcal{F}_n]=S_n^2-Var(S_n)$.

I tried to do this by induction, but I'm already having trouble with the base case:

n=1: $E[S_2^2-Var(S_2)|\mathcal{F}_1]=E[X_1^2|\mathcal{F}_1]+2E[X_1|\mathcal{F}_1]E[X_2 \mathcal{F}_1]+E[X_2^2|\mathcal{F}_1]-E[E(X_1^2)+E(X_2^2)|\mathcal{F}_1]=...?$

I would be glad, if you could tell me, if I did something wrong so far, or, if not, how to continue.

Thanks for your help!


Here I will use $E_n[\cdot]$ for $E[\cdot | \mathcal{F}_n]$.Note that since the $X_i$ are independent, you have $E_n(\sum_1^N X_i)^2 = \sum_1^N E_n X_i^2$, all of the cross terms vanish since for $i < j$ we have by iterated conditioning $$E_nX_iX_j = E_nE_i X_iX_j = E_n X_i E_iX_j = E_n X_i E X_j = E_n X_i 0 = 0.$$ Use this and the fact variances add for independent random variables as well.

| cite | improve this answer | |

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.