Roulette

Q: Which bet has the lowest house edge?

On a European wheel, all bets have the same 2.70% house edge . Single-number, color, dozen, column, split, street, corner — mathematically identical long-run cost per dollar wagered. The 'difference' between bets is variance : a single-number bet at 35:1 has very high variance (you'll lose 36 spins in a row before winning one big), while red/black has low variance (about half spins win). On a French wheel with La Partage, even-money bets drop to 1.35% — making them the lowest-edge bet at the casino, period. On American wheels, the only 'worse' bet than 5.26% is the 'five-number bet' (0, 00, 1, 2, 3) at 7.89% — never take it.

/ruːˈlɛt/ · French: "little wheel"

European roulette table with green felt — the canonical wheel layout — *Image: Pixabay Content License.*

I. Origin — Pascal, the Wheel, and the Pursuit of Perpetual Motion

The roulette wheel was invented (in essence) in 1655 by Blaise Pascal as a byproduct of his attempt to design a perpetual-motion machine. The bidirectional spinning rotor — wheel one way, ball the opposite — turned out to be unsuitable for energy generation but perfectly suited to gambling. By 1796 a recognizable roulette table with the modern wheel layout was operating at the Palais-Royal in Paris.

The original wheel had both 0 and 00 — identical to today's American wheel. In 1843, the brothers Louis and François Blanc removed the 00 to create the modern European wheel for the German spa-casino at Bad Homburg, halving the house edge and attracting players away from rival casinos. The Blancs later founded Monte Carlo's Casino de Monte-Carlo in 1863, exporting their single-zero standard across Europe.

The double-zero wheel migrated to the United States with French emigrants during the early 19th-century Mississippi-River casino era, becoming the American standard.

II. The Wheel — Geometry of the Pockets

Roulette wheel with numbered pockets — the precision rotor of modern casino math — *Image: Pixabay Content License.*

The numbering on the wheel is not sequential. It is engineered to maximize randomness from any physical wheel imperfection:

Red/black alternation — every other pocket alternates color (excepting 0)
Odd/even balance — high odd and high even numbers are distributed across the wheel
Sum of adjacent pairs — opposite pockets typically sum to 37 or 38 (on European/American respectively)
Diametrically opposite are always one odd + one even

This sequence is the same on all standardized casino wheels:

European (single-zero): 0, 32, 15, 19, 4, 21, 2, 25, 17, 34, 6, 27, 13, 36, 11, 30, 8, 23, 10, 5, 24, 16, 33, 1, 20, 14, 31, 9, 22, 18, 29, 7, 28, 12, 35, 3, 26
American (double-zero): 0, 28, 9, 26, 30, 11, 7, 20, 32, 17, 5, 22, 34, 15, 3, 24, 36, 13, 1, 00, 27, 10, 25, 29, 12, 8, 19, 31, 18, 6, 21, 33, 16, 4, 23, 35, 14, 2

III. The Betting Layout — Inside and Outside

Bet Type	Numbers Covered	Payout	House Edge (Euro)
Straight up (single)	1	35:1	2.70%
Split	2 adjacent	17:1	2.70%
Street (3 in a row)	3	11:1	2.70%
Corner / Square	4	8:1	2.70%
Line / Six-number	6	5:1	2.70%
Dozen (1-12 / 13-24 / 25-36)	12	2:1	2.70%
Column	12	2:1	2.70%
Red / Black	18	1:1	2.70%
Odd / Even	18	1:1	2.70%
High (19-36) / Low (1-18)	18	1:1	2.70%
Even-money + La Partage	18	1:1 (lose half on 0)	1.35%

The single house edge (2.70%) appears in every bet because the 0 pocket "steals" the same fraction of expected value across all betting structures. Mathematicians call this the "fair-payoff deficit" — straight-up pays 35:1 but the true odds are 36:1 (37 pockets minus the winner). The 1-pocket difference is the casino's 2.70%.

IV. The Math — Why 2.70% Is the Same on Every Bet

Consider straight-up on European wheel:

Win probability: 1/37 ≈ 0.0270
Pay on win: 35 × bet
Lose probability: 36/37 ≈ 0.973
Expected value: (1/37)(35) + (36/37)(−1) = (35 − 36)/37 = −1/37 = −2.70%

Consider red:

Win probability: 18/37 ≈ 0.4865
Pay on win: 1 × bet
Lose probability: 19/37 ≈ 0.5135
Expected value: (18/37)(1) + (19/37)(−1) = −1/37 = −2.70%

The −1/37 fraction is invariant. It is the percentage of all spins that land on 0, scaled by the unit-bet loss. Every roulette bet is the same long-run trade.

V. French Roulette — La Partage and En Prison

French roulette tables (predominantly at Monte Carlo, Casino Barrière chain, some London Mayfair clubs) include two player-favorable rules on even-money bets when 0 hits:

La Partage ("the sharing"): Player loses half stake on 0. House edge on red/black, odd/even, high/low = 1.35%.
En Prison: Player's stake is held for one more spin. Win on the second spin returns the original stake (no profit); loss = total loss. Mathematically equivalent house edge to La Partage when applied identically.

These rules halve the cost of even-money bets — making French roulette one of the lowest-edge bets in any casino, narrowly beaten only by basic-strategy blackjack and certain video-poker pay tables.

VI. Famous Strategies and Why They All Fail

Casino chips on roulette betting table — every system meets the same 2.70% math — *Image: Pixabay Content License.*

Martingale: Double after every loss on even-money bets. Mathematically: works until you hit the table maximum (typically $5,000 on a $10 table = 9 doublings) or your bankroll. Probability of 9 consecutive losses on red ≈ 0.5135⁹ ≈ 0.0017 — happens once every 600 spins. When it does, you lose $5,110 trying to recover $10.
D'Alembert: Increase by 1 unit after loss, decrease by 1 after win. Smoother bankroll curve, but identical long-run house edge.
Fibonacci: Bet according to Fibonacci sequence. Same 2.70%.
Labouchère / Cancellation: Cross off numbers after wins, add after losses. Same 2.70%.

The pattern: no betting system can change the house edge, because each spin is independent and 2.70% is built into every wager. The only mathematical winning strategy is biased-wheel exploitation (now extinct on modern wheels) or computer prediction (now criminalized).

VII. The Biased-Wheel Era (1873-1991)

Three documented million-dollar exploitations of mechanical wheel bias:

Joseph Jagger (1873, Monte Carlo) — Yorkshire engineer paid six clerks to record every spin on six wheels for several weeks. Identified wheel 4 had a slight tilt favoring 9 numbers. Bet those numbers exclusively, winning £65,000 (≈ $7.5M in today's value). The casino restored balance by rotating wheels nightly.
Billy Walters (1986, Atlantic City) — Tracked thousands of spins, identified bias at the Golden Nugget, won approximately $3.8M before casinos noticed and changed wheels.
Gonzalo Garcia-Pelayo (1992, Madrid Casino Gran Madrid) — Logged 24,000+ spins per wheel, fed data to a statistical program; identified six biased wheels. Won approximately €1.5M over five years. Subsequently sued for "cheating" but Spanish courts ruled the technique legal.

By 2000, manufactured wheels achieved tolerances under 0.05mm; biased-wheel exploitation became essentially impossible. The casino's wheelscan software flags any 1.5%+ pocket deviation across 5,000 spins for immediate maintenance.

VIII. Online Roulette and Live-Dealer Variants

Online roulette splits into two domains:

RNG roulette — each spin a fresh PRNG sample; mathematically equivalent to a perfect physical wheel; pace 400-600 spins/hour. Speed is the casino's hidden weapon — at 600 spins/hour with $10 bets, expected loss on European is $162/hour; on American $315/hour.
Live-dealer roulette — Evolution Gaming's "Immersive Roulette" and "Lightning Roulette" dominate. Real wheel, real dealer, multi-camera angle, streamed from Riga / Manila / Bucharest. Lightning Roulette adds random multipliers (50x-500x) on selected straight-up numbers, attractive marketing but worse mathematics (house edge rises to 2.93% on single-number bets, 3.10% on multipliers).

IX. Famous Roulette Spins in History

August 18, 1913 — Monte Carlo "Black 26": Black hit 26 consecutive times on a single wheel. Players lost millions betting red, convinced "red is due." Each spin remained independent at 48.6% red probability. The textbook gambler's fallacy demonstration.
2004 — Ashley Revell at Plaza Las Vegas: 32-year-old British man sold everything he owned ($135,300) and bet it all on red. Won. Walked away with $270,600.
2016 — Pedro Grendene Bartelle, Hotel Conrad Punta del Este: Brazilian businessman bet $100,000 on straight-up 32 (his lucky number). Won $3.5M at 35:1.

X. FAQ · Sources · Responsibility

European vs American roulette — how big is the difference?

Massive. European roulette has 37 pockets (0-36) with a 2.70% house edge on every bet. American roulette adds a 00 pocket (38 total) which raises the house edge to 5.26% on every bet — almost double. Over 100 spins of $10 single-number bets, the expected loss is $27 on European, $52.60 on American. Always play European wheels when given a choice. Monte Carlo, London, most of Europe = European. Las Vegas Strip = mostly American (a handful of single-zero tables exist at Bellagio, Wynn, and some boutique rooms — usually with higher minimums).

What is 'En Prison' and 'La Partage'?

Two player-favorable rules found on French roulette tables (single-zero wheels): ① La Partage ('the sharing') — when the ball lands on 0 and you've bet on an even-money proposition (red/black, odd/even, high/low), you lose only half your stake. ② En Prison ('imprisoned') — when 0 hits on an even-money bet, your bet is 'imprisoned' for one more spin; if it wins, you get the original stake back (no profit); if it loses, you lose everything. Both rules cut the house edge on even-money bets from 2.70% to 1.35%. Look for these rules in French/Monte Carlo casinos — they are rare in the US.

Does the 'gambler's fallacy' apply to roulette?

Yes — and it's the most expensive mental error in the game. Every spin is independent. A roulette wheel that has hit red 10 times in a row has exactly the same probability (18/37 ≈ 48.6%) of hitting red on the 11th spin. The wheel has no memory. The most famous demonstration: the 1913 Monte Carlo 'black 26' incident — black hit 26 consecutive times. Players lost millions betting on red, convinced 'red is due.' Each spin remained independent at 48.6% red probability.

Can biased wheels actually be exploited?

Historically yes; modern era almost never. Pre-1980s wheels were hand-assembled with measurable mechanical biases — certain pockets had slightly elevated frets, or the rotor wobbled imperceptibly. Joseph Jagger (1873, Monte Carlo) and Billy Walters / Gonzalo Garcia-Pelayo (1992, Madrid) won millions by tracking thousands of spins to identify biased pockets. Modern manufactured wheels (Cammegh, John Huxley, TCSJohnHuxley) are precision-machined to tolerances under 0.05mm, with regular auditing. The casino's tracking software (Cammegh Wheelscan, eDeck Roulette Vision) flags any pocket with > 1.5% probability deviation across 5,000 spins. Practical bias exploitation in 2026 is essentially extinct on regulated wheels.

What about computer prediction of the ball's landing?

Theoretically possible; practically banned. The 1970s 'Eudaemons' team led by Doyne Farmer and Norman Packard built shoe-computers timing the ball's deceleration to predict octant landing with ~44% advantage over the house. They won approximately $10,000 before mechanical failures and casino vigilance ended the project. Modern jurisdictions (Nevada, UK, Macau) explicitly criminalize the use of any device to predict roulette outcomes (Nevada NRS 465.075 — felony, up to 10 years). The legal prosecution effectively closed this advantage technique, regardless of mathematical feasibility.

Which bet has the lowest house edge?

On a European wheel, all bets have the same 2.70% house edge. Single-number, color, dozen, column, split, street, corner — mathematically identical long-run cost per dollar wagered. The 'difference' between bets is variance: a single-number bet at 35:1 has very high variance (you'll lose 36 spins in a row before winning one big), while red/black has low variance (about half spins win). On a French wheel with La Partage, even-money bets drop to 1.35% — making them the lowest-edge bet at the casino, period. On American wheels, the only 'worse' bet than 5.26% is the 'five-number bet' (0, 00, 1, 2, 3) at 7.89% — never take it.

Sources

Edward O. Thorp (1969), Optimal Gambling Systems for Favorable Games, Review of the International Statistical Institute
Cammegh Wheelscan Technical Documentation, 2023 Edition
Nevada Revised Statutes §465.075, Use of Device to Determine Outcomes
Gonzalo Garcia-Pelayo (2009), La Fabulosa Historia de los Pelayos, RBA Libros
Casino de Monte-Carlo, Histoire d'une institution: 1863-2013

Responsible play: This article is historical and mathematical analysis, not gambling advice. The 2.70% European edge looks small, but at typical session length (200+ spins) translates to meaningful losses. If gambling stops being entertainment, contact the National Council on Problem Gambling: 1-800-522-4700.