The fair games theorem (optional stopping)
ProvedFamousTheorems.submartingale_iff_expected_stoppedValue_monoThe fair games theorem, in the form of Doob's optional stopping theorem.
Let be an adapted, integrable process on a filtered probability space. Then is a submartingale if and only if for every pair of bounded stopping times ,
Read left to right this says you cannot beat an unfavourable game by clever timing: no stopping rule based only on information available so far can improve the expected payoff. Read right to left it says this "no strategy helps" property actually characterises submartingales — the two formulations carry the same information. For a martingale both inequalities hold, so the expectation is constant across all bounded stopping times, which is the precise sense in which a fair game stays fair.
The boundedness hypothesis is essential rather than technical. Without it the gambler's doubling strategy — keep doubling the stake until the first win — is a stopping time that turns a fair game into a sure profit, which is why casinos impose table limits and why the unbounded case needs uniform integrability.
Doob developed the theory in the 1940s and 50s; optional stopping is the tool behind the classical ruin problems, Wald's identity, and much of mathematical finance, where it is the reason no-arbitrage pricing works.
Formalization note. Filtration ℕ m0 is the increasing family of -algebras, stopping
times take values in ℕ∞, and stoppedValue f τ is . The
result is Mathlib's MeasureTheory.submartingale_iff_expected_stoppedValue_mono.
import Mathlib
namespace FamousTheorems
open MeasureTheory ProbabilityTheory Filter Set intervalIntegral
open scoped Real Topology ENNReal
theorem submartingale_iff_expected_stoppedValue_mono {Ω : Type*} {m0 : MeasurableSpace Ω}
{μ : Measure Ω} {𝒢 : Filtration ℕ m0} {f : ℕ → Ω → ℝ} [SigmaFiniteFiltration μ 𝒢]
(hadp : StronglyAdapted 𝒢 f) (hint : ∀ i, Integrable (f i) μ) :
Submartingale f 𝒢 μ ↔ ∀ τ σ : Ω → ℕ∞, IsStoppingTime 𝒢 τ → IsStoppingTime 𝒢 σ →
τ ≤ σ → (∃ N : ℕ, ∀ x, σ x ≤ N) → μ[stoppedValue f τ] ≤ μ[stoppedValue f σ] := by sorry
end FamousTheorems