Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The Original Question: Demonstrate that every martingale is a local martingale.

Attempt at a Solution:

Consider the standard setup of this problem: $\mathscr{F}_t$ is the filtration that satisfies the normal conditions, $(\sigma_n)_{n\in \mathbb{N}}$ is the monotone increasing sequence of stopping times with $\text{lim}_{n \to \infty}\sigma_n = \infty$ and $\{X_t\}_{t \in [0,\infty]}$ is a martingale.

To show that $X$ is a local martingale I will show that $E[X_\infty | \mathscr{F}_t] = X_{min(t , \sigma_n)}$.

We know that $X_\infty = X_{min(\infty,\sigma_n)}=X_{\sigma_n}$ because $\sigma_n \in \mathbb{R}_+$. It then follows that:

$E[X_{\sigma_n}|\mathscr{F}_t] = X_{\sigma_n} = X_{min(t,\sigma_n)}$.

However this incorrect for any $t < \sigma_n$, and so $\exists$ $\geq 1$ mistakes in my attempted solution.

MY Question: Am I correct? If not, what's wrong with my solution, and what's some good hints to a correct solution (or just post the correct solution).

share|cite|improve this question
You just have to show the existence of one particular localization $(\sigma_n)_{n\in\mathbb{N}}$ such that $(X^{\sigma_n}_t)_{t\geq 0}$ is a martingale. It seems to me you're on the way to showing that $(X^{\sigma_n}_t)_{t\geq 0}$ is a martingale for every localization $(\sigma_n)_{n\in\mathbb{N}}$ (which does not hold). – Stefan Hansen Nov 8 '12 at 17:26
@StefanHansen Could you please dumb down your criticism of my attempt? I wikipedia'd localization but couldn't relate it to what you're saying. – Jase Nov 8 '12 at 17:45
Sorry, I think the correct term is localizing sequence, see e.g. this. It's just a term meaning any sequence $(\sigma_n)_{n\in\mathbb{N}}$ of stopping times increasing to $\infty$, which is used to define local martingales (and other local properties). – Stefan Hansen Nov 8 '12 at 18:00
@StefanHansen I think I don't understand what you mean by "one particular" sequence $(\sigma_n)_{n \in \mathbb{N}}$. Guessing what you mean; we can specify that each $\sigma_n = \text{inf}\{t \geq 0 : X_t = n\}$ where $X_{min(t,\sigma_n)}$ is the stopped process. However I can't see how this can help me solve the problem. – Jase Nov 9 '12 at 5:13

This is too long to be a comment, so I posted it as an answer instead. Let me first introduce the definitions I use of a martingale, local martingale and localizing sequence.

A sequence $(\sigma_n)_{n\in\mathbb{N}}$ is called a localizing sequence if $\sigma_n$ is a stopping time for all $n\in\mathbb{N}$ and $\sigma_n\uparrow \infty$ a.s.

A stochastic process $(X_t)_{t\geq 0}$ is called a martingale (wrt. $(\mathscr{F}_t)_{t\geq 0}$) if it is integrable, adapted and satisfies $$ E[X_t\mid\mathscr{F}_s]=X_s \quad \text{a.s.} $$ for every $0\leq s<t$.

A stochastic process $(X_t)_{t\geq 0}$ is called a local martingale if there exists a localizing sequence $(\sigma_n)_{n\in\mathbb{N}}$ such that the stopped process $(X_t^{\sigma_n})_{t\geq 0}=(X_{t\wedge \sigma_n})_{t\geq 0}$ is a martingale for every $n\in\mathbb{N}$.

Now, you're starting with a martingale $(X_t)_{t\geq 0}$ and in order to show that it is also a local martingale, you just have to find a localizing sequence $(\sigma_n)_{n\in\mathbb{N}}$ such that $(X_{t\wedge\sigma_n})_{t\geq 0}$ is a martingale. Your task is now to find such sequence using the fact that $(X_t)_{t\geq 0}$ already is a martingale. To this end, note that a stopping time is a mapping from $\Omega$ into $\mathbb{R}_+\cup \{\infty\}$ and not into $\mathbb{R}_+$.

share|cite|improve this answer
Hi I think that by taking $\sigma_n=n$ as a localizing sequence, it is clear from the fact that a stopped martingale by a bounded stopping time stays a martinagle, that every martingales are local martingales. Best regards – TheBridge Nov 9 '12 at 13:14
You could also just pick $\sigma_n=\infty$ for all $n$. – Stefan Hansen Nov 9 '12 at 15:24

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.