Roulette house edge is the casino’s long-run average advantage on the amount wagered. It comes from a simple pricing gap: the wheel contains zero pockets that count as losing outcomes for many bets, but the standard payouts are set as if those extra losing pockets were not fully compensated.
That price depends first on the wheel and rules. Standard single-zero roulette is about 2.70% on ordinary wagers. Standard double-zero roulette is about 5.26% on most ordinary wagers. French half-back rules can reduce the effective edge on even-money bets further.
The house edge begins with a payout gap
Imagine a hypothetical roulette wheel containing only the numbers 1 through 36. A straight-up wager on one number would have 35 losing outcomes for every winning outcome, so a 35-to-1 payout would be fair.
Real single-zero roulette adds a 37th pocket: zero. The casino still pays the straight-up winner 35 to 1. The player now has 36 losing pockets but is compensated as though there were only 35.
That one-pocket mismatch is the single-zero house advantage:
$$House\ Edge=\frac{1}{37}\approx 2.7027%$$
Double-zero roulette adds another green pocket while leaving the familiar payouts largely unchanged. For ordinary bets the disadvantage becomes:
$$House\ Edge=\frac{2}{38}\approx 5.2632%$$
The Wizard of Odds roulette basics gives the standard comparison and the common exceptions. Official rules such as the Nevada roulette rules of play and Massachusetts roulette rules show why the exact wheel and settlement rule must be identified before quoting an edge.
Derive the single-zero edge from an even-money bet
A red bet on a standard European-style single-zero wheel wins on 18 red pockets. It loses on 18 black pockets plus zero.
With a one-unit stake:
$$EV=\left(\frac{18}{37}\times 1\right)-\left(\frac{19}{37}\times 1\right)=-\frac{1}{37}$$
The player loses an average of 1/37 of the stake over long repetition, so the house edge is 1/37, or about 2.70%.
This is the same percentage produced by a standard straight-up bet, split, street, corner, six-line, dozen, or column on the same single-zero wheel. The hit rates and payout sizes differ, but the standard pricing gap is arranged to produce the same rate.
That does not mean every possible roulette wager everywhere has the same edge. Side bets, promotional variants, unusual baskets, and rule changes must be priced separately.
Double zero almost doubles the ordinary price
On a standard double-zero wheel, an even-money wager still has 18 winning numbered pockets. It now loses on 18 opposite-color pockets, zero, and double zero: 20 losing pockets out of 38.
$$EV=\left(\frac{18}{38}\times 1\right)-\left(\frac{20}{38}\times 1\right)=-\frac{2}{38}$$
That is about -5.26% of the stake.
For a player choosing between otherwise comparable tables, this difference is substantial. If $10,000 of ordinary action is wagered over time:
- at 2.70%, theoretical loss is about $270;
- at 5.26%, theoretical loss is about $526.
Actual results can be wildly different over a short sample, but the pricing difference is real before the first spin occurs.
This is why wheel selection usually matters more than arguments over whether red, a dozen, or a single number is “best.”
The important exception: not every American wager is 5.26%
A useful correction to the slogan “all roulette bets have the same edge” is the traditional American five-number basket covering 0, 00, 1, 2, and 3. At the common 6-to-1 payout, its house edge is about 7.89%, worse than the ordinary double-zero rate.
The lesson is broader than that one wager: price the exact bet. Standard outside and inside bets on a given conventional wheel usually share the standard wheel edge, but special combinations can break the pattern.
For wager-by-wager probabilities, use roulette odds and why most roulette bets have the same house edge rather than turning a rule of thumb into an absolute law.
La Partage changes settlement, so it changes the edge
French-style rules can improve even-money wagers because zero no longer necessarily causes the full stake to disappear.
Under La Partage, when zero lands on an eligible even-money wager, the player receives half the stake back. That reduces the effective edge to roughly 1.35% on those bets on a single-zero wheel.
An En Prison rule can create a similar economic result under the classic French treatment, although the exact handling of repeated zeros and imprisoned bets can depend on house rules. The important point is that the settlement changed.
This is a genuine mathematical improvement because one outcome that previously lost the whole unit now loses less. A betting progression does not do that.
If those rules are available, compare them before worrying about a favorite number or system.
House edge, expected value, RTP, variance, and hold are not synonyms
Roulette discussions often mix five different metrics.
House edge is the expected casino advantage as a percentage of action.
Player expected value is the same pricing relationship viewed from the player side, usually negative and expressible in units or dollars.
RTP is the expected return percentage: on a standard single-zero wager, roughly 97.30% under the usual convention.
Variance describes the spread of possible results around the average. Straight-up bets have much larger swings than red/black even when their standard house-edge percentage is the same.
Hold is an operational casino result often measured against buy-in/drop over a period. It can differ greatly from theoretical edge because chips recycle, session lengths vary, and short-term luck is noisy.
Keeping these terms separate prevents a common mistake: seeing a table hold 20% for a shift and concluding the game has a 20% house edge. It does not.
For the player-side pricing formula, use roulette expected value. For swing size, use roulette variance.
Total action turns a small percentage into money
The house edge is applied to amount wagered, not simply to the amount originally exchanged for chips.
A player can buy in for $200, wager $20, win, wager the returned chips again, and continue cycling the same bankroll. After 50 spins at $20 per spin, planned action is:
$$50\times $20=$1{,}000$$
On standard single-zero roulette, long-run expected loss attached to that action is approximately:
$$$1{,}000\times 2.7027%\approx $27.03$$
On standard double-zero roulette:
$$$1{,}000\times 5.2632%\approx $52.63$$
The player might actually be ahead after 50 spins. The calculation is not forecasting the session. It is showing what repeated action costs on average.
This is why pace matters. A small bet repeated rapidly can create more expected cost than a larger bet placed only a few times.
Why the casino can rely on edge without controlling the next spin
Roulette profitability does not require the operator to know or manipulate the next number. The pricing advantage is built into the approved wheel and paytable.
At table level, management cares about handle, game speed, staffing, player ratings, disputes, and actual win versus theoretical win. Surveillance cares about procedure, wheel integrity, late bets, payout accuracy, and suspicious activity. None of those functions requires predicting the next pocket.
Across large amounts of action, the known mathematical price becomes useful for budgeting and performance analysis. A single table can win or lose heavily during one shift; the portfolio does not need every short sample to match theory exactly.
That operational distinction is central to understanding why roulette can be random at the spin level and still be commercially predictable at scale.
The best player-controlled lever is usually the rule set
You cannot choose where the ball lands. You often can choose which roulette offer to accept.
A practical hierarchy is:
- Prefer a properly run single-zero wheel over double zero when the rest of the conditions are comparable.
- If French half-back rules are available on even-money wagers, understand whether they apply and how zero is handled.
- Check special bets separately instead of assuming the standard wheel edge.
- Keep the stake and number of spins aligned with the session budget.
- Do not mistake lower variance for lower edge or a progression for better pricing.
The “best roulette bet” question therefore begins with the game price, not the layout pattern.
Translate the percentage into a session cost before you play
A percentage becomes meaningful only when attached to action. If you expect to wager $2,500 in total, the rough theoretical cost is about $67.57 at 2.7027% and about $131.58 at 5.2632%.
That does not tell you where the session will finish. It tells you what price you are accepting for the amount of roulette you plan to buy.
Continue with European roulette house edge and American roulette house edge for wheel-specific detail. Use the house edge calculator or expected loss calculator to convert a rate into money, and read why roulette is easy to understand but hard to beat if the next temptation is a staking system rather than a better rule set.