A case study: how a small change in roulette rules shifts expected loss

In any casino, roulette looks deceptively stable: the wheel spins, the ball drops, and payouts appear fixed. Yet expected loss is driven not by drama but by arithmetic. This case study shows how a minor rule tweak can materially change the player’s long-run cost, even when the game feels identical. We focus on European roulette (single zero) and a common rule variation applied to even-money bets: “en prison” or “la partage”.

Under standard European rules, there are 37 pockets (1–36 plus 0). An even-money bet (red/black, odd/even, high/low) wins with probability 18/37, loses with probability 18/37, and lands on zero with probability 1/37. With a £1 stake, the expected value is (18/37)(+£1) + (18/37)(-£1) + (1/37)(-£1) = -£1/37, an expected loss of about 2.70% per spin. Now apply “la partage”: if zero appears, you lose only half your stake. The expectation becomes (18/37)(+£1) + (18/37)(-£1) + (1/37)(-£0.50) = -£0.50/37, or roughly 1.35%. That single concession halves the house edge on even-money bets, a shift large enough to change bankroll longevity and strategy evaluation. For comparison tools and rule summaries used by players, see Casoola.

Professional analysts in iGaming often stress that “small print” dominates outcomes. A well-known voice is Michael “Wizard” Shackleford, who popularised rigorous expected-value breakdowns and educated players on how rule variants alter the edge; his primary social profile is WizardOfOdds on X. Broader scrutiny of the sector also matters: The New York Times report on the rise of sports betting and addiction concerns illustrates how policy and product design shape consumer harm—an argument that applies equally to roulette rule tweaks that quietly shift expected loss.