How to Win at Roulette: 11 Tips for a Cheaper Table
Roulette pays the same way whatever you back. Red, a corner or a single number: on a single-zero wheel the expected cost per unit staked is identical, so most advice about how to win at roulette spends its time sorting bets that were never different from each other. The three choices that do move your result are the wheel you sit at, the rule that wheel runs and how much turnover you push across it.
Every bet on a single-zero wheel costs the same 2.70%
A straight-up number pays 35 to 1 on a wheel with 37 pockets. Win one spin in 37 and collect 35, lose the other 36, and you are down 1 unit for every 37 staked. That is 2.70%.
Run the identical sum on red, on a corner, on a street or on a dozen and it lands on 2.70% every time, because every payout on the layout is set the same single step below true odds. The figures below are computed from the published pocket count and payout table, because there is no return figure to use.
Evolution publishes no RTP on any of its roulette game pages, so a roulette comparison either runs on house edge derived from the rules or on numbers somebody else made up. Slots are the opposite case: a printed return figure on the game page is the whole reason the highest RTP casino games can be ranked on a studio's own number.
What bet choice does change is how violently the result swings. The right-hand column is the standard deviation of the result per unit staked, the number that decides how long your stake survives at a fixed bet size.
| Bet | Pays | House edge | Swing per unit |
|---|---|---|---|
| Red, black, odd, even, high, low | 1 to 1 | 2.70% | 1.00 |
| Dozen or column | 2 to 1 | 2.70% | 1.40 |
| Line, six numbers | 5 to 1 | 2.70% | 2.21 |
| Corner, four numbers | 8 to 1 | 2.70% | 2.79 |
| Street, three numbers | 11 to 1 | 2.70% | 3.28 |
| Split, two numbers | 17 to 1 | 2.70% | 4.07 |
| Straight up, one number | 35 to 1 | 2.70% | 5.84 |
Pick the wheel and the rule before you pick a bet
1. Count the zeros before you place a chip
A single-zero wheel has 37 pockets and charges 2.70% of everything staked. Add a second green pocket and the same-looking layout charges 5.26%. Per 1,000 units staked that is an expected 27.03 against 52.63, close to double the cost for a difference you can see from across the room. Pocket count is the first of the four lines that tell the best roulette variants apart, and the only one you can check without opening anything.
2. Take La Partage where the table offers it
La Partage resolves the zero immediately: you back an even-money box, zero lands, half your stake comes straight back and the round is over. The sum becomes 18 wins in 37, 18 full losses and one half loss, which is 0.5 divided by 37, or 1.35%. Evolution describes the rule on its live roulette hub. It halves the edge on even-money bets only, so a corner on that same table still costs 2.70%. At 1.35% those boxes are the cheapest bet in roulette, and still nearly triple the roughly 0.5% that basic strategy reaches on the best blackjack games.
3. Check whether the table runs En Prison instead
En Prison reaches a similar number by a different mechanism. Zero lands and your whole stake is locked for one more spin rather than returned; win that spin and you get the stake back with nothing on top. Where a second zero re-imprisons the bet the edge computes to 1.35%, and where a second zero takes it the edge is about 1.39%. Either way your money is committed to a spin you did not choose to make.
4. Never take the five-number bet
On a double-zero layout, the bet covering 0, 00, 1, 2 and 3 pays 6 to 1. Five pockets in 38 collect 30 units and lose 33, which is 3 units per 38 staked, or 7.89%. Every other bet on that wheel charges 5.26%, so this one costs exactly half again as much. It is the only bet on either wheel with its own worse number.
| Wheel and rule | Pockets | Even-money bets | Every other bet |
|---|---|---|---|
| Single-zero, no special rule | 37 | 2.70% | 2.70% |
| Single-zero, La Partage | 37 | 1.35% | 2.70% |
| Single-zero, En Prison | 37 | 1.35% to 1.39% | 2.70% |
| Double-zero | 38 | 5.26% | 5.26%, five-number bet 7.89% |
What a multiplier variant charges you for the multiplier
5. Read the straight-up line before the multiplier headline
Evolution's Lightning Roulette page publishes the entire trade in two lines. One to five lucky numbers per round carry multiplied payouts of between 50x and 500x, and non-multiplied straight-up bets pay 29 to 1. That 29 is where the multipliers are funded from.
Price the 29 to 1 payout on its own, as though the lightning never found your number: one win in 37 collecting 29 against 36 losses is 7 units down per 37 staked, an 18.92% edge. The multipliers have to hand back 16.22 points of that on average. Evolution publishes neither how often each multiplier value lands nor the average lucky-number count per round, so the straight-up return is not published and cannot be derived from anything that is.
The headline 500x is paid for out of your ordinary wins. Every straight-up win the lightning misses returns 29 units where a standard wheel would have paid 35, a win roughly 17% smaller, and that shortfall is the budget the multipliers are drawn from.
6. Bet anywhere except the straight-up and the trade never touches you
Evolution states that splits, corners, red and black and dozens all pay the same as standard roulette on those tables. The wheel has one zero, so every one of those bets carries the plain 2.70% while only the straight-up is funding the show. The same trade turns up across the best game show games, where a headline multiplier is always paid for somewhere in the base paytable.
The bigger the ceiling, the more base payout funds it. Evolution's XXXtreme Lightning Roulette page publishes multipliers to 2,000x and up to ten lucky bet spots, but no straight-up payout at all, which makes it the first line to go find in the paytable. Rules separate one live table from the next far more than lobby art does, so read them before you pick an operator from the live casino ranking.
Shape the swing and control the turnover
7. Choose your bet from the swing column
A straight-up number swings about 5.8 times as hard per unit staked as red does, for exactly the same expected cost. That is the whole content of the inside-versus-outside argument: identical price, different ride. Pick the swing figure that matches the number of rounds you want out of a fixed stake.
8. Watch the clock on the table you pick
Evolution publishes 25 seconds a round for Speed Roulette. Assuming you play those rounds back to back with no breaks, that is 144 rounds an hour, and at 1 unit a round the expected cost is 3.89 units an hour on a single-zero wheel against 7.58 on a double-zero one. A slower table at the same stake charges less per hour for the same edge, because less turnover crosses it.
The edge is a rate charged on turnover. Halving your rounds per hour cuts your expected cost per hour just as precisely as halving your stake does, and pace is the cheaper of the two levers to pull.
What no staking pattern can do
9. Refuse progressions, and check the arithmetic yourself
A conference paper hosted in a university research repository derives the Martingale distribution in full and concludes that a long-run loss is inevitable because of the negative expected value. The one-line reason: the table charges 2.70% of everything that crosses it, so expected loss is 2.70% of total turnover, and a doubling sequence exists to raise total turnover.
Run one seven-step sequence on a single-zero wheel: 1, 2, 4, 8, 16, 32 and 64 is 127 units risked to win 1, and only the run that loses all seven ever stakes all 127. That run comes up 0.94% of the time, about 1 in 106 sequences. Most sequences stop at the first win, so the average one pushes 7.59 units of turnover and costs 0.21 units, while the 127 units of turnover a full run pushes cost 3.43 units. The 3.43 belongs to the turnover, not to the sequence, and raising turnover is the only thing the doubling achieves.
10. Ignore the scoreboard above the table
Peer-reviewed research on observed casino roulette play found that after a streak of five or more identical outcomes, players are significantly more likely to bet against the streak than with it. The wheel does not read its own history board. A number that has landed five times running has the same 1 in 37 chance on the next spin as the coldest number on the layout, so the board is decoration and betting from it costs the same 2.70% as anything else.
What a contribution rate does to the cheapest bet
11. Read the contribution line before you claim an offer
Offers commonly exclude even-money betting outright, which removes exactly the bet La Partage made cheapest, and count whatever roulette turnover is left at a fraction of face value. A 10% contribution rate means 1 unit staked clears 0.10 of the requirement, so the same wagering multiple takes ten times the turnover to finish. Contribution rates are the first line worth reading in the terms behind any offer on the bonus ranking, because they decide whether roulette turnover counts at all.
Key takeaways
- Count the zeros first. A single-zero wheel charges 2.70% of turnover and a double-zero wheel charges 5.26%.
- On a single-zero wheel every bet carries the same edge, so bet choice sets your swing and nothing else.
- La Partage halves the even-money edge to 1.35% by returning half your stake at once. En Prison locks the stake for a spin instead and lands at 1.35% to 1.39%.
- The five-number bet on a double-zero layout costs 7.89% against 5.26% for everything else.
- Multiplier variants fund the multiplier by cutting the straight-up payout to 29 to 1, and leave every other bet at 2.70%.
- Expected loss is a rate on turnover, so rounds per hour and stake size move it and staking patterns do not.
Frequently asked questions
Can you actually win at roulette?
You can win a session, but you cannot lift the expected return above 1.00 per unit staked. Computed from the published pocket count and payouts, every bet on a single-zero wheel returns 0.973 per unit, and no order or size of bets changes that figure. Winning at roulette means paying the lowest available rate on the least turnover you can get away with.
What is the best bet in roulette?
On a single-zero wheel there is no best bet, because all of them carry the same 2.70% house edge computed from the published payouts. Red and a straight-up number cost the same per unit staked and differ only in how hard the result swings. On a double-zero wheel the useful answer is negative: the five-number bet is the worst at 7.89% against 5.26% for everything else.
Is single-zero roulette really better than double-zero?
Yes, by 2.56 percentage points, computed from the pocket count and payouts each layout publishes. A single-zero wheel charges 2.70% of turnover and a double-zero wheel charges 5.26%. Per 1,000 units staked that is an expected 27.03 against 52.63, on two tables that look almost identical.
What does La Partage do to the house edge?
It halves the edge on even-money bets to 1.35%. When zero lands, half your stake is returned immediately and the round ends, so that pocket costs 0.5 units instead of 1. Computed over the published rule, 0.5 divided by 37 is 1.35%, and it applies to even-money bets only, so a corner on the same table still costs 2.70%.
Is En Prison the same as La Partage?
No, and the mechanisms differ. La Partage hands back half your stake at once, while En Prison locks the whole stake for one more spin and returns it only if that spin wins. Computed from the published rule, En Prison lands at 1.35% where a second zero re-imprisons the bet and about 1.39% where a second zero takes it.
Why does the five-number bet cost more than every other bet?
Because its payout sits further below true odds than any other bet on the layout. Five pockets in 38 paying 6 to 1 collect 30 units and lose 33, which is 3 units per 38 staked, or 7.89% computed from the published payout. Every other bet on that wheel costs 5.26%, so the five-number bet charges exactly half again as much.
Does Lightning Roulette pay better than standard roulette?
Evolution publishes no return figure for it, so there is no published answer to give. What the studio does publish is that non-multiplied straight-up bets pay 29 to 1 rather than 35 to 1, and that every other bet pays the same as standard roulette. Computed from that payout alone, a 29 to 1 straight-up with no multiplier would carry an 18.92% edge, which is the gap the multipliers refill.
Does the Martingale system work?
No. Expected loss is a fixed rate on total turnover, so raising your stake after a loss raises turnover and raises expected loss along with it. A seven-step doubling sequence risks 127 units to win 1, but it only stakes all 127 on the run that loses every spin, which comes up about 1 in 106 sequences on a single-zero wheel. Computed at 2.70% from the published pocket count, those 127 units of turnover cost 3.43 units, while the average sequence stops at its first win, pushes 7.59 units of turnover and costs 0.21. The 3.43 is the price of the turnover rather than the price of running the pattern once.
Do hot and cold numbers mean anything?
No, because each spin is independent of the last. A number that has just landed five times has the same 1 in 37 chance on the next spin as one that has not landed all evening. Peer-reviewed research on observed casino roulette play found players bet against a streak of five or more far more often than with it, which is a pattern in the players rather than in the wheel.
How much does an hour of roulette cost?
It depends entirely on rounds per hour and stake size, because the edge is a rate charged on turnover. Evolution publishes 25 seconds per round for Speed Roulette, which computes to 144 rounds an hour if you play them back to back with no breaks. At 1 unit a round that is an expected 3.89 units an hour on a single-zero wheel and 7.58 on a double-zero one.
Which roulette bet has the lowest variance?
Red, black, odd, even, high and low, at a swing of about 1.00 per unit staked, computed as the standard deviation of the result per unit. A straight-up number swings about 5.8 times as hard for exactly the same 2.70% expected cost. Dropping the swing lengthens the session at a fixed stake without changing what it costs on average.