Roulette
Roulette is the cleanest house edge in the building, so this table shows it to you instead of hiding it. Place any bet and you get its exact win probability, payout, and expected value before the wheel turns. Then simulate a few thousand spins and watch your realized return walk down to the theoretical number and stay there.
Recent:
Your bets, priced
Run it a few thousand times
Repeats the layout you've placed and settles every spin instantly. Small samples wander; large ones don't.
Simulated spins move the statistics above but not the bankroll. Staking $15 a spin for ten thousand spins would bust a $1,000 stack long before the average had room to show up — that's gambler's ruin, which is a different lesson than this one.
Spin log
Why every bet has the same edge
A bet covering n pockets on a wheel of N wins with probability n/N and pays m to 1. Its expected value per unit staked is
EV = (n/N)·m − (1 − n/N)
The payouts are set as if there were only 36 pockets — a straight-up number pays 35:1, a dozen pays 2:1, red pays 1:1, and in each case n·(m+1) = 36. So EV = 36/N − 1, which is −1/37 = −2.70% on a single-zero wheel and −2/38 = −5.26% on a double-zero wheel. The zeros are the entire business model: every bet is paid at 36-pocket odds on a 37- or 38-pocket wheel.
Which is why no betting pattern helps. Combining, doubling, chasing, or spreading across the felt just mixes bets that all carry the same percentage, and the sum of many −2.70% bets is a −2.70% bet. The only lever that matters is which wheel you sit at — and the American layout has one extra pocket that costs you nearly double.
The one bet worse than the rest: the American five-number bet on 0-00-1-2-3, which pays 6:1 where the geometry calls for 6.6:1, running −7.89%. It isn't on this layout on purpose.