Roulette Odds

Key Takeaways

• Roulette odds are pure counting: 37 pockets on European wheels, 38 on American — every probability in the game follows from that one number.
• Payouts are built for a fictional 36-pocket wheel, so the zero(s) create the house edge: 2.70% European, 5.26% American, on every standard bet.
• Within one wheel, all standard bets cost the same per dollar wagered — bet choice changes rhythm and variance, never expected value.
• The two real decisions with mathematical content: wheel selection (European over American) and rule selection (la partage / en prison halves even-money edge to 1.35%).
• No progression, pattern or "due number" system alters any of this; spins are independent and the wheel has no memory.

How Roulette Odds Work

A roulette wheel is an equal-probability device: each pocket — numbers 1–36 plus a green 0 (and 00 on American wheels) — has an identical chance per spin. Every bet is just a subset of pockets, so its probability is:

P(bet wins) = pockets covered / total pockets

Payouts, however, are calibrated to a wheel with only the 36 numbers: a straight bet pays 35:1, exactly fair if there were 36 pockets. The green pockets are outside that payment scheme — they're where the casino's margin physically lives on the wheel.

The full odds table

European (single-zero, 37 pockets):

American (double-zero, 38 pockets): every probability shrinks — straight 2.63%, even-money 47.37%, dozens 31.58% — and the edge doubles to 5.26% across the board, with one exception: the five-number basket (0-00-1-2-3) pays 6:1 for a 13.16% chance, an edge of 7.89%, the worst standard bet in the game.

The edge calculation, shown once for the European straight bet:

EV = (1/37) × 35 − (36/37) × 1 = −1/37 ≈ −2.70% of stake

Run the same arithmetic on any row of the table and −1/37 appears again: the uniformity of the edge is roulette's defining mathematical fact. The casino doesn't care which bets you make on a given wheel — only how much and for how long.

Green: the most-asked question

The green pockets get outsized search attention, so precisely: on a European wheel, green means the single 0 — 1/37 = 2.70%, playable as an ordinary straight bet at 35:1 with the standard 2.70% edge. On an American wheel, "hitting green" (0 or 00) is 2/38 = 5.26%; the dedicated 0-00 split pays 17:1, again at the standard 5.26% edge. Betting green is neither better nor worse than any other pocket — its mystique is color, not mathematics.

French rules: the real odds improvement

Two rule variants, historically French, apply when the ball lands on zero while you hold an even-money bet:

• La partage — half the stake is returned immediately.
• En prison — the stake is "imprisoned" for the next spin; winning releases it.

Both cut the even-money edge from 2.70% to 1.35% — the cheapest bet available on any standard roulette table, and cheaper than most slots by a wide margin. The improvement applies only to even-money bets; straights and dozens stay at 2.70%.

The wheel-choice table

Choosing French over American rules cuts the price of identical play by nearly 75% — the single largest "strategy" gain available in roulette, and it happens before the first chip is placed. Online, wheel selection is trivially available: the SlotCodex catalog tracks 115 roulette games from 55 providers (July 2026 snapshot), spanning European, American, French and novelty formats.

Called bets: the racetrack, priced

European tables often carry a "racetrack" for announced (called) bets — fixed chip packages covering wheel sectors rather than layout geometry. They look esoteric; they're just bundles of the standard bets, and they inherit the standard 2.70% edge exactly:

Because each package is assembled from splits, straights and corners, its expected value is the sum of its parts: −2.70% of total chips, always. Sector bets change where on the wheel your variance lives — a Voisins player experiences the game as "zero's neighborhood vs the rest" — but no arrangement of chips rearranges the arithmetic. The racetrack is ergonomics, not edge.

A worked session

Concrete expectation math for a typical evening — 100 spins at $10 per spin ($1,000 turnover):

The standard deviation on 100 even-money bets (~$100 on $10 stakes) dwarfs the expected loss — which is precisely why sessions feel winnable and often are. The edge only becomes visible in aggregates: across a thousand such sessions, the averages converge on the table above with statistical certainty. Both facts are true at once; confusing them is the engine of most roulette folklore.

Why systems don't work

Every roulette system — Martingale, Fibonacci, D'Alembert, hot numbers, sector tracking on RNG games — collides with two facts:

1. Independence. Each spin's distribution is identical and unaffected by history. Ten reds in a row leave P(red) at exactly 18/37; the "law of averages" operates by dilution over millions of trials, not by short-term correction.
2. Linearity of expectation. Any combination of negative-EV bets, in any sequence or sizing, sums to negative EV. Martingale redistributes outcomes — many small wins, rare catastrophic losses at the table cap — without moving the mean a cent.

What systems do change is the shape of your session: variance, duration, and the probability of small-win sessions versus ruin. That is a legitimate preference space; just price it honestly — every shape costs the same 2.70% (or 5.26%) of turnover.

A historical footnote for completeness: biased physical wheels (worn frets favoring sectors) were genuinely exploited in the 19th–20th centuries, and dealer-signature/ballistics claims persist around live wheels. Modern casino maintenance and RNG-driven online roulette leave no practical residue of either. Novelty variants — multipliers on random numbers ("lightning"-style) — rebalance payouts around the same or slightly higher total edge; their published RTPs, typically just above 97%, are the honest comparison figure.

For Players

• Pick the wheel first. European over American always; French rules over both when available. This one choice dominates everything else in the game combined.
• Skip the basket. The American 0-00-1-2-3 bet at 7.89% is the only standard bet strictly worse than its table-mates.
• Choose bets for texture, not value. Straights: rare 35:1 hits, high variance. Even-money: near-coinflip rhythm. Dozens: in between. All identically priced per wheel — pick the session shape you enjoy.
• Budget by turnover. Expected cost = total wagered × edge. $1 spins for 200 spins on a European wheel: about $5.40 expected loss; on American, $10.52. Variance around that is large, which is the entertainment.
• Online, check the format label. "Roulette" titles vary: auto-roulette, live wheels, RNG, multiplier variants — the info screen's RTP (97.30% for standard European) identifies the underlying math instantly.

For the Industry

• Format mix is a margin decision with churn consequences. American-wheel prevalence in some markets doubles hold per wager but is increasingly visible to players as odds literacy spreads; several online lobbies now lead with European/French tables as an acquisition signal.
• Multiplier variants trade transparency for excitement. Lightning-style games publish slightly lower even-money RTPs in exchange for jackpot texture; disclosure quality (per-bet RTP in the info screen) is the compliance frontier regulators actually check.
• RNG vs live is a certification split. RNG roulette is certified like a slot (RNG + paytable model); live wheels fall under physical-equipment and studio standards. Catalogs should model them as distinct products even when branded identically — SlotCodex tracks them under one game type with source-level distinctions.
• Odds content is high-intent SEO. Roulette-odds queries dominate casino-math search volume (the largest single cluster in our 2026 Ahrefs encyclopedia export); accurate, computation-backed odds tables outperform generic reviews for both players and answer engines.