Definition:Martingale/Discrete Time/Definition 1

Definition

Let $\struct {\Omega, \Sigma, \sequence {\FF_n}_{n \mathop \ge 0}, \Pr}$ be a filtered probability space.

Let $\sequence {X_n}_{n \mathop \ge 0}$ be an adapted stochastic process.

We say that $\sequence {X_n}_{n \mathop \in \N}$ is a $\sequence {\FF_n}_{n \ge 0}$-martingale if and only if:

$(1): \quad$ $X_n$ is integrable for each $n \in \Z_{\ge 0}$

$(2): \quad \forall n \ge 0: \expect {X_{n + 1} \mid \FF_n} = X_n$.

Equation $(2)$ is understood as follows:

for any version $\expect {X_{n + 1} \mid \FF_n}$ of the conditional expectation of $X_{n + 1}$ given $\FF_n$, we have:

$\expect {X_{n + 1} \mid \FF_n} = X_n$ almost surely.

Also see

• Equivalence of Definitions of Martingale in Discrete Time

• Results about martingales can be found here.

Sources

• 1991: David Williams: Probability with Martingales ... (previous) ... (next): $10.2$: Adapted process