Auto Roulette RTP, House Edge & Payouts: The 97.30% Explained (Live Data)
By LiveCasinoData editorial team · Updated 2026-08-28 · 14 min read
Evolution's Auto Roulette returns 97.30% on every bet spot, because the single green zero gives the house a 2.70% edge on all of them. There are no multipliers, side bets or bonus rounds to shift that figure. Across 103,221 spins tracked over the last 90 days, every number landed between 2.62% and 2.78% against a theoretical 2.70%, and zero came up 2.73% of the time.
Key takeaways
- Every bet on Auto Roulette has the same theoretical RTP of 97.30% and the same 2.70% house edge; there is no higher-RTP spot to find.
- The edge comes from one pocket: a straight-up win pays 35:1 on a 1-in-37 event that would need 36:1 to break even.
- Bets differ only in variance. A straight-up bet wins 2.70% of spins at 35:1, an even-money bet wins 48.65% of spins at 1:1.
- Over 103,221 tracked spins, the hottest number (10) hit 2.78% and the coldest (21) hit 2.62%, a gap of 167 hits that random variation fully explains.
- At about 70 spins an hour, the 2.70% edge costs an expected 18.9 units per hour for every 10 units staked per spin, whichever bet you choose.
Auto Roulette returns 97.30% of stakes to players in the long run, on every single bet, and keeps 2.70%. That is the arithmetic of a single-zero wheel: 37 pockets, payouts priced as if there were 36. Nothing else in the game moves the number, because there are no multipliers, no bonus round and no side bets. Our tracker has logged 275,852 spins of Evolution's Auto Roulette since 2025-12-28, and the last 103,221 of them show a wheel that behaves exactly as the 97.30% figure says it should.
What is the RTP of Auto Roulette?
The theoretical RTP is 97.30%, published by Evolution in the game rules and identical for every bet spot on the layout. The house edge is 100% minus that, 2.70%, which is exactly 1 divided by 37.
RTP describes the long-run share of wagered money that flows back to players. If a million units were bet on red over enough spins, roughly 973,000 would be returned as winnings and 27,000 would stay with the house. The same ratio holds for a million units on number 17, on the first dozen, or on a Voisins du Zero call bet.
This uniformity separates Auto Roulette from most of the games we track. Game shows have a different RTP per bet spot, and lightning-style roulette pays reduced straight-up odds to fund multipliers. Auto Roulette has one wheel, one payout table and one edge. The automation changes nothing: a compressed-air launcher fires the ball instead of a croupier, and the ball still has 37 equal pockets to fall into.
Where does the 2.70% house edge come from?
From a single pocket. There are 37 numbers on the wheel, so a straight-up bet wins one time in 37. A fair payout for that would be 36:1 (36 units profit plus your stake back). The table pays 35:1. That missing unit, spread over 37 outcomes, is the edge.
Bet 1 unit on a number 37 times and you hit once on average: 36 units lost on the misses, 35 collected on the hit, minus 1 unit over 37 spins, or 2.70% per spin.
Every other bet is built from the same shortfall. An even-money bet on red covers 18 pockets and pays 1:1. Fair odds would be 19:18, since 19 pockets (18 black plus zero) beat you. Over 37 spins you win 18 and lose 19, net minus 1, again 1/37 per spin. A dozen covers 12 pockets at 2:1: 12 wins of 2 units against 25 losses of 1 unit, net minus 1 over 37 spins. The pattern never breaks.
The only rule that would change this is la partage or en prison, where zero returns half of an even-money stake. That lifts even-money RTP to 98.65%. The standard Auto Roulette table does not use it; zero takes everything. Evolution offers separate French-rules tables, and operators label them differently in the lobby.
What does every Auto Roulette bet pay?
All bets pay at odds that sit one unit short of fair on a 37-pocket wheel. Here is the complete list with win probability and expected return.
| Bet | Numbers covered | Payout | Win probability | RTP | House edge |
|---|---|---|---|---|---|
| Straight up | 1 | 35:1 | 2.70% | 97.30% | 2.70% |
| Split | 2 | 17:1 | 5.41% | 97.30% | 2.70% |
| Street | 3 | 11:1 | 8.11% | 97.30% | 2.70% |
| Corner | 4 | 8:1 | 10.81% | 97.30% | 2.70% |
| Line (six) | 6 | 5:1 | 16.22% | 97.30% | 2.70% |
| Dozen | 12 | 2:1 | 32.43% | 97.30% | 2.70% |
| Column | 12 | 2:1 | 32.43% | 97.30% | 2.70% |
| Red / Black | 18 | 1:1 | 48.65% | 97.30% | 2.70% |
| Odd / Even | 18 | 1:1 | 48.65% | 97.30% | 2.70% |
| 1-18 / 19-36 | 18 | 1:1 | 48.65% | 97.30% | 2.70% |
Multiply any row's win probability by its payout plus one and you get 0.973, the RTP as a decimal.
Racetrack bets follow the same logic because they are bundles of ordinary bets. Voisins du Zero places 9 chips across 17 numbers using splits, a street and corners. Tiers du Cylindre uses 6 chips on 12 numbers as splits. Orphelins uses 5 chips on 8 numbers, and Jeu Zero uses 4 chips on 7 numbers. Each component chip returns 97.30%, so the bundle does too. A sector bet does not buy a better price; it buys a specific slice of the wheel, and the streaks and sectors page covers how often each slice has actually landed.
Which bet has the highest RTP in Auto Roulette?
No bet is higher than another. Straight up, red, first dozen and a 9-chip Voisins all cost 2.70% per unit staked, so the search for a best-value spot ends before it starts.
Where bets do differ is in how the 97.30% arrives. On red, you get back roughly 97 units per 100 staked in a steady drip: win two units nearly half the time, lose one slightly more than half the time. On a single number, you get back the same 97 units per 100 on average, but as 36 units of return concentrated into fewer than 3 spins per 100, with the other 97 spins paying nothing.
That difference is variance, and it is the only real choice the layout offers. Our strategy page ranks combinations by coverage and variance rather than by RTP, because RTP would rank them all in a tie.
How volatile is each bet?
Variance per spin runs from almost nothing on even-money bets to large on straight-up bets, even though expected loss is identical. The table shows the standard deviation of the result per unit staked, computed from each bet's payout and probability.
| Bet | Win probability | Result if win | Result if lose | Standard deviation per unit |
|---|---|---|---|---|
| Straight up | 2.70% | +35 | -1 | 5.84 |
| Split | 5.41% | +17 | -1 | 4.07 |
| Street | 8.11% | +11 | -1 | 3.28 |
| Corner | 10.81% | +8 | -1 | 2.80 |
| Line | 16.22% | +5 | -1 | 2.21 |
| Dozen / Column | 32.43% | +2 | -1 | 1.40 |
| Even-money | 48.65% | +1 | -1 | 1.00 |
The expected result per unit is minus 0.027 in every row. The standard deviation is what changes, by a factor of nearly six between red and a single number.
In practice this means an hour of 70 spins at 10 units on red has an expected loss of about 19 units with a typical swing of roughly plus or minus 84 units around it. The same hour on a single number has the same expected loss of 19 units, but a typical swing of about plus or minus 490 units, and the most likely outcome of any given hour is that the number does not hit at all. The probability that a chosen number misses 70 straight spins is (36/37)^70, about 14.7%.
Streaks make outside bets feel more volatile than the numbers suggest. Our 90-day data holds a run of 16 reds and a run of 13 blacks; a player flat-betting black through the 16-red run lost 16 units in about 13 minutes. The probability of any spin starting a 16-run of one colour is about 0.0021%, which over 103,000 spins produces one or two such runs on average.
What is the maximum win in Auto Roulette?
The largest payout on any single bet is 35:1 on a straight-up number. There are no multipliers, so a 100-unit stake on 17 pays 3,500 profit when it hits and never more. The tracker's multiplier field for this game is fixed at 1.00 for exactly that reason.
Above the payout ratio sits a per-round cap of 500,000 in the account currency, with the exact figure set by the operator. Reaching it on Auto Roulette requires a very large stake spread over many spots; for nearly all players it never comes into play.
Bet limits vary by operator. Minimums are typically 0.10 or 0.20 per spot, with maximums in the thousands. Limits do not touch RTP: a 0.10 bet and a 1,000 bet on the same spot have the same expected return per unit. They do matter for progression systems, because a doubling sequence hits the table maximum within a handful of losses, and the strategy page works through that arithmetic.
What does the house edge cost per hour?
About 18.9 units for every 10 units staked per spin. The calculation is stake times spins times 2.70%, and the number of spins is the part players underestimate on an automated wheel.
Auto Roulette runs at roughly one spin every 51 seconds by our feed timing, about 70 spins an hour and 1,690 a day. No dealer pauses to greet players, and the betting window reopens the moment the previous result is settled. A session that feels short can easily contain 150 spins.
| Stake per spin | Spins per hour | Total wagered | Expected loss per hour |
|---|---|---|---|
| 1 | 70 | 70 | 1.89 |
| 5 | 70 | 350 | 9.45 |
| 10 | 70 | 700 | 18.90 |
| 50 | 70 | 3,500 | 94.50 |
| 100 | 70 | 7,000 | 189.00 |
The stake column means total chips on the layout per spin, not per spot. Covering 20 numbers with 1 unit each is a 20-unit stake and costs 0.54 units per spin in expectation. Speed of play is the largest lever on cost in this game, larger than any choice of bet, and sitting out spins reduces expected loss in direct proportion.
Does the observed number distribution match the theoretical 2.70%?
Yes. Over the last 90 days, 103,221 spins, every one of the 37 numbers landed between 2.62% and 2.78% of the time. Theory says 2.70% each, which is 2,790 hits per number in a sample this size.
| Measure | Theory | Observed (90 days) |
|---|---|---|
| Most frequent number | 2.70% | 10 at 2.78% (2,870 hits) |
| Least frequent number | 2.70% | 21 at 2.62% (2,703 hits) |
| Red | 48.65% | 48.63% |
| Black | 48.65% | 48.64% |
| Zero | 2.70% | 2.73% |
| 1-18 | 48.65% | 48.67% |
| Odd | 48.65% | 48.59% |
| 1st / 2nd / 3rd dozen | 32.43% each | 32.49% / 32.29% / 32.49% |
| Column 1 / 2 / 3 | 32.43% each | 32.55% / 32.41% / 32.31% |
| Voisins du Zero (17 numbers) | 45.95% | 45.62% |
| Tiers du Cylindre (12) | 32.43% | 32.61% |
| Orphelins (8) | 21.62% | 21.77% |
The gap between the hottest and coldest number is 167 hits across 103,000 spins. For a fair wheel, the number of hits on any pocket has a standard deviation of about 52 in this sample, so the top number sits roughly 1.5 standard deviations above expectation and the bottom one about 1.7 below. With 37 numbers in play, extremes of that size are the normal spread, not evidence of a lean.
The same holds for repeats and zero gaps. The same number followed itself 2,716 times, 2.6% of spins against a theoretical 2.70%. The average gap between zeros was 36.6 spins against a theoretical 37. Three consecutive zeros, a 1-in-50,653 event per starting spin, occurred once in the window, which is close to the two occurrences you would expect per 100,000 spins.
One caveat on the sample: the collector was offline from 2026-06-30 to 2026-07-09. That reduces the number of spins in the window but does not bias the shares, since the wheel does not know whether we are watching. Spins-per-day figures for early July are affected; percentages are not.
Why does a matching distribution not mean a matching RTP for you?
Because RTP is a long-run average and your session is a short sample. The 90-day figures confirm the wheel is unbiased, which is the precondition for the 97.30% figure being true. They do not mean that any player experienced 97.30% over their own spins.
Consider the zero drought that ended on 2026-08-23. The wheel went 396 spins without a zero, against an average gap of 37, so a player betting even-money through those 5.7 hours never once lost to the green pocket. Two further droughts of 277 and 270 spins both ended on 2026-07-25. In the other direction, the same 90 days contain a 469-spin absence of number 29, during which anyone flat-betting 29 lost every unit they placed.
Neither stretch changed the next spin's odds. The chance of zero on the spin after a 396-spin drought was 2.70%, the same as the spin after a zero. This is the point at which the live tracker is most often misread: a drought is a record of the past, and the wheel's memory of it is nil.
Our RTP monitor compares observed and certified figures across the games we track; for Auto Roulette the observed row is essentially the hit-share table above, since no multipliers exist to move the return.
How do you read the Auto Roulette figures on our tracker?
The live stats page shows number and colour shares, sector heat, streaks and the zero drought, all computed from the same round records summarised here. A few reading notes save the most common misunderstandings.
Hot and cold numbers are ranked by share over the selected window. In a 24-hour window of about 1,690 spins, each number is expected to hit about 46 times with a standard deviation near 6.7, so the top number will routinely show 60 hits and the bottom one 33. Those are the tails of a fair distribution, not a signal, and the ranking reshuffles daily. Over 90 days the spread narrows to the 2.62% to 2.78% range above; over a longer window it would narrow further.
Colour and dozen shares are the quickest sanity check on the wheel. Values within a few tenths of a point of 48.65% and 32.43% over tens of thousands of spins are what an unbiased wheel produces; a persistent gap of a full point would be worth a closer look, and we have never seen one.
Sector heat maps the Voisins, Tiers and Orphelins shares. Our 90-day figures (45.62%, 32.61%, 21.77%) sit within a third of a point of theory. The sectors are large groups of numbers, so their shares settle faster than single-number shares.
The zero drought and streak counters are descriptive. They answer "how long has it been" and "what was the longest run", and nothing about what comes next.
Is Auto Roulette's RTP the same as other Evolution roulette tables?
For the standard single-zero tables, yes, 97.30% is the common figure, but the games we track differ in how they reach it. Speed Roulette and Immersive Roulette are dealer tables with the same wheel and payout table; they differ in pace and presentation, not in edge. Speed Auto Roulette is a faster automated table with its own feed and is not covered by this tracker.
Lightning Roulette and its XXXtreme sibling keep 97.30% on outside bets but pay 29:1 on straight-up numbers to fund random multipliers, which changes both the straight-up RTP and, far more, its variance. Auto Roulette has no equivalent trade-off. The 35:1 payout is the full payout, every spin, which makes it the simplest roulette table Evolution runs.
A 2.70% edge is one of the lower prices in live casino, and it is still a price, paid on every spin, that no bet selection reduces. Set a stake and a spin count before you start, use the pacing tools on our observatory if sessions run long, and read the full Auto Roulette guide for the wider picture of the game beyond its maths.
Frequently asked questions
What is the RTP of Evolution Auto Roulette?
Auto Roulette has a theoretical RTP of 97.30% on every bet, which Evolution publishes in the game rules. The wheel is a standard single-zero European wheel with 37 pockets, and all payouts (35:1 straight up, 17:1 split, 2:1 dozens, 1:1 red or black) are set so that the zero pocket keeps 1/37 of stakes. There are no multipliers or side bets that pay at a different rate.
Does Auto Roulette have la partage or en prison?
The standard Auto Roulette table tracked here does not. When zero lands, every outside bet loses in full. Evolution runs separate tables with French rules, sometimes labelled La Partage, where even-money bets get half back on zero and the RTP on those bets rises to 98.65%. Those tables have different names in the lobby and are not the same game as this one.
Which bet has the best odds in Auto Roulette?
None, if odds means expected return. Straight up, split, street, corner, line, dozen, column and even-money bets all return 97.30% in the long run. If odds means chance of winning a given spin, red or black wins 48.65% of the time and a single number 2.70%. The trade is frequency for payout size; the house edge is identical either way.
How much does Auto Roulette cost to play per hour?
The wheel spins about 70 times an hour. Expected loss is 2.70% of everything staked, so 10 units on every spin costs about 18.9 units an hour on average, 100 units per spin costs 189. The actual result in any hour is dominated by variance, especially on inside bets, but the average cost scales with stake times spins, not with which bet you choose.
Is Auto Roulette rigged or biased?
Our tracker gives no sign of it. Over 103,221 spins in 90 days, all 37 numbers landed within 0.08 percentage points of the 2.70% theory, red came up 48.63% and black 48.64%, and zero 2.73%. A biased pocket would show up as a persistent excess over that many spins. The same-number repeat rate of 2.6% and the zero gap of 36.6 spins also sit where an unbiased wheel puts them.