Biased Roulette Wheels: Why It Does Not Work

The idea that a physically imperfect roulette wheel can be exploited for profit sounds compelling — but modern casino maintenance has made true wheel bias an essentially extinct opportunity.

Wheel bias is not a myth in the historical sense. There are documented cases of players winning meaningful sums by tracking which pockets a wheel favored over thousands of spins. What is largely a myth is the notion that the same approach is available to you today at any reputable casino. Understanding why requires a look at both the physics and the math.

What Wheel Bias Actually Means

A biased wheel is one where manufacturing imperfections, worn frets (the metal dividers between pockets), a slightly unlevel surface, or a loose ball track cause certain numbers to appear more often than the 1-in-37 (European) or 1-in-38 (American) probability the game is built on.

On a perfectly balanced European wheel, every number carries the same 2.70% house edge. If a wheel were genuinely biased — say, number 17 landed 1 in 30 spins instead of 1 in 37 — a player betting only on 17 at the standard 35:1 payout would have a positive expected value on that particular wager. That reversal of the house edge is the entire premise of bias play.

The edge gained in the most favorable historical cases was roughly 2–5 percentage points above expectation. That sounds attractive until you factor in everything required to find and exploit it.

The Historical Case: The Eudaemons

The most cited example of bias play comes from the Eudaemons, a group of physics graduate students in the late 1970s who used concealed computers to predict where a roulette ball would land based on the speed of the wheel and ball at the moment of release. Their edge was real, and they did win money. But their work was primarily about physics prediction, not traditional bias tracking — and it required custom hardware that is now illegal to use in virtually every jurisdiction.

The classic bias-tracking method — simply recording outcomes by hand until certain numbers show statistical deviation — is older still, dating to the nineteenth century. Joseph Jagger famously hired clerks to log results at Monte Carlo in 1873 and reportedly exploited a genuinely defective wheel. These cases share a common thread: the wheels of that era were machined by hand to tolerances that modern CNC manufacturing renders obsolete.

Modern Casino Wheel Maintenance

Today’s major casinos treat their roulette wheels as precision instruments. A few things happen routinely:

Tolerance checks: Manufacturers like Cammegh and TCSJohnHuxley publish wheel specifications with pocket-size variance held below 0.1%. At that tolerance, the statistical advantage from any natural bias would be smaller than the noise in any realistic sample size.
Level verification: Wheels are checked for horizontal alignment regularly. Even a minor tilt can create a very slight bias — so casinos monitor it and correct it.
Rotation schedules: Wheels are physically swapped between tables on regular schedules. A player investing days or weeks logging outcomes at one table may arrive the next morning to find a different wheel in place.
Ball and fret replacement: The ball and the frets wear over time. Rather than allowing wear to accumulate into a detectable bias, casinos replace components on preventive schedules.
Electronic monitoring: Many modern wheels include sensors that log where the ball lands. Surveillance systems can flag statistically unusual distributions far faster than any human observer — often before a player could gather enough data to act on it.

The Sample Size Problem

Even if a wheel were slightly biased, detecting it with statistical confidence requires a very large number of observations. To distinguish a genuine 1-in-30 bias on a single number from normal variance on a 1-in-37 wheel, you need roughly 10,000 to 30,000 recorded spins at a minimum — and that estimate assumes the bias is consistent and clean. In practice, you would need more.

A busy roulette table in a casino might handle 40–50 spins per hour. At that rate, 10,000 spins represents over 200 hours of observation at a single table. During that time, the wheel will almost certainly have been rotated, serviced, or replaced. The casino’s own monitoring systems would have flagged any anomaly long before your dataset was large enough to act on.

Even in the unlikely scenario where a biased wheel stayed in place long enough to identify, you would then need to win back the time investment of hundreds of hours of logging with a 2–5% edge in your favor — an edge that only applies to the specific biased number, not to any other bet you place while sitting at the table.

Online Roulette: No Physical Wheel at All

If you play roulette at an online casino, the question of wheel bias does not apply. Online roulette uses a random number generator (RNG) — a software algorithm that produces outcomes independently each spin. There is no physical ball, no frets, no levelness to worry about, and no bias to detect. The math of the game is exactly what it says it is: 2.70% on a European wheel, 5.26% on an American wheel.

For a deeper look at how RNGs work and how they are certified, see our article on roulette RNG fairness.

What This Means for Your Play

Wheel bias is a fascinating chapter in gambling history, not a practical strategy for the present. The correct framework for playing roulette is the one the game actually offers: a known, fixed house edge that applies equally to every spin, every bet, and every player.

If you want to minimize what you give up to the house, choosing a European wheel over an American one is the most straightforward decision available — dropping the house edge from 5.26% to 2.70% by avoiding the double-zero. Beyond that, the game is what it is: a negative-expectation bet where the entertainment is the point.

For a broader look at roulette variants and their edge differences, or to explore how the game works, those resources will give you a clearer picture of what you are actually playing.

Frequently Asked

Has wheel bias ever been successfully exploited in a real casino?

Yes, historically. The most documented cases come from the nineteenth and early twentieth centuries, when wheels were less precisely manufactured. The Eudaemons in the 1970s used physics prediction with concealed computers — a different method that is now illegal. Genuine bias exploitation at modern casinos is extremely rare and essentially unverified.

How many spins would I need to detect bias?

A rough statistical rule of thumb is 10,000 to 30,000 spins per number to reach meaningful confidence that an apparent deviation is real bias rather than normal variance. At 40–50 spins per hour, that is hundreds of hours of observation before you could act on the data — and casinos rotate wheels on schedules that would almost certainly interrupt that window.

Does bias apply to online roulette?

No. Online roulette uses a software RNG with no physical components. There is nothing to be biased. The outcomes are determined by certified algorithms, not a spinning wheel and a bouncing ball.

What is the best strategy for roulette given that bias is not viable?

Choose European (single-zero) over American (double-zero) wherever available. That single decision cuts the house edge from 5.26% to 2.70%. Beyond that, all bets on a standard wheel carry the same edge — there is no bet selection strategy that improves your expected outcome. See our roulette strategy overview for more detail.

Sources & Further Reading

Ethier, S. N. The Doctrine of Chances: Probabilistic Aspects of Gambling (Springer, 2010) — Chapter 17 covers wheel bias and the statistical requirements for detection.
Bass, T. The Eudaemonic Pie (Houghton Mifflin, 1985) — Narrative account of the Eudaemons’ physics-prediction project.
Wizard of Odds — Roulette — house edge figures and rules summaries for all major variants.

Educational explanation only. No real-money gambling happens on LearnTheOdds.

Responsible gambling: Play for entertainment, not income — the math favors the house over time. Set limits, never chase losses, and if it stops being fun, take a break. 21+. Need help? Call 1-800-MY-RESET (1800myreset.org).