The Wheel Decides the Odds Before You Choose a Bet
Most roulette questions start with the felt, but the arithmetic starts with the wheel. A European wheel has 37 pockets; an American wheel has 38. That single pocket difference is the most important number on the table, because every bet you can place is priced against those pockets, not against the payout the table advertises.
When a straight-up bet pays 35 to 1, the payout is short of what a fair game would pay on a 37-pocket wheel. A fair game would pay 36 to 1. The missing unit is the house edge, and it is present on every bet, every spin, regardless of what you wagered before or what you wager next.
This means the question "how to bet roulette odds" is really two questions. The first is mechanical: what does each bet pay, and what is the probability behind it. The second is financial: how much of each wager does the house keep on average, and what does that do to a session. Both answers come from the same arithmetic.
Reading a Roulette Bet as Expected Value
Expected value is the average result of a bet if it were repeated a very large number of times. It is not a prediction of the next spin. To build it, multiply each possible outcome by its probability, then add the results.
As an example, suppose you place 1 unit on a single number on a European wheel. The probability of winning is 1 in 37, and the probability of losing is 36 in 37. The expected value is (1/37 × 35) plus (36/37 × -1). That arithmetic produces a negative number, and that negative number is the house edge expressed per unit wagered.
The same calculation works for every bet on the layout. A split bet covering two numbers pays 17 to 1, but the probability of winning is 2 in 37. A street bet covering three numbers pays 11 to 1, but the probability is 3 in 37. The payout rises as the probability falls, but the gap between the two never closes in the player's favour.
How the Payouts and Probabilities Line Up
The table below compares the main roulette bet types on a European wheel. The payout column is what the table advertises. The probability column is the chance of winning on a single spin. The expected loss column is what the arithmetic produces per unit wagered, before any variation in a real session.
| Bet type | Numbers covered | Payout | Probability of winning | Expected loss per unit |
|---|---|---|---|---|
| Straight-up | 1 | 35 to 1 | 1 in 37 | Same as the house edge |
| Split | 2 | 17 to 1 | 2 in 37 | Same as the house edge |
| Street | 3 | 11 to 1 | 3 in 37 | Same as the house edge |
| Corner | 4 | 8 to 1 | 4 in 37 | Same as the house edge |
| Six line | 6 | 5 to 1 | 6 in 37 | Same as the house edge |
| Dozen/column | 12 | 2 to 1 | 12 in 37 | Same as the house edge |
| Red/black, odd/even, high/low | 18 | 1 to 1 | 18 in 37 | Same as the house edge |
The striking column is the last one. On a European wheel, every bet listed carries the same expected loss per unit. The bet type changes the shape of the session, not the long-run cost. A player who spreads chips across red and a corner is not diversifying risk in the way an investor diversifies a portfolio; they are placing multiple bets, each with the same negative expectation, and the total exposure is the sum of the chips on the felt.
What the American Wheel Changes
An American wheel has 38 pockets: the numbers 1 through 36, a single zero, and a double zero. The payouts are the same as on a European wheel. A straight-up bet still pays 35 to 1, a split still pays 17 to 1, and red/black still pays even money.
The probabilities are not the same. A straight-up bet now wins 1 time in 38, not 1 in 37. The payout has not increased to compensate. The result is a larger gap between the fair payout and the advertised payout, which means a larger house edge on every bet.
This is the cleanest example of how the odds are built into the wheel rather than chosen by the player. The player's decision is which bet to place, but the wheel's pocket count is fixed before the first chip lands. On an American layout, the same bet costs more in expected loss per unit than it does on a European layout, because the extra pocket adds a losing outcome without adding a winning one.
RTP and House Edge Are Complements Under One Rule Set
Return to Player (RTP) is a theoretical percentage of total wagered money a game returns to players over a very large number of spins or hands. It is a long-run statistical average, not a per-session guarantee. House edge is the mathematical advantage the operator holds on a given game, expressed as a percentage of each wager. House edge and RTP are complements: they sum to 100% for the same game.
The phrase "for the same game" matters. A European wheel and an American wheel are different games with different RTPs, because the pocket count changes the underlying probabilities. A bet on red and a bet on a single number on the same wheel share the same RTP, because the expected loss per unit is identical. RTP is a property of the game and its rule set, not of the bet selection.
This is why a player cannot improve the RTP by changing bet type on the same wheel. They can change the variance, the frequency of winning spins, and the size of the payouts, but the long-run return is fixed by the wheel. The only bet-selection change that alters RTP on roulette is moving between wheel types, where the rules of the game itself change.
Why a Short Session Looks Nothing Like the Average
Expected value describes a very large number of repetitions. A session of fifty or a hundred spins is not a very large number. In a short sample, the actual result can be far above or far below the expected value, and that gap is variance.
Variance is how payouts are distributed over time. A straight-up bet has high variance: most spins lose the entire stake, and a rare spin returns 35 units. An even-money bet has lower variance: roughly half the spins return a unit, and the losing spins are smaller in relative terms. Both bets carry the same expected loss per unit on the same wheel, but they produce very different session shapes.
A player who hits a straight-up number in the first ten spins is not beating the house edge. They are experiencing the high-variance tail of a negative-expectation game. The edge is still there, applied to every spin, including the winning one. The arithmetic does not pause because the player is ahead.
Decisions Per Hour Change Total Exposure, Not the Edge
Roulette is a game of repeated independent trials. Each spin is a new wager with the same expected loss per unit. The total expected loss of a session is the expected loss per unit multiplied by the total amount wagered, not by the amount the player brought to the table.
Total amount wagered is a product of three things: the stake per spin, the number of spins per hour, and the number of hours played. A player betting 5 units per spin at 40 spins per hour is wagering 200 units per hour. The same player at 60 spins per hour is wagering 300 units per hour, and the expected loss rises proportionally.
This is the lever a player controls. Bet selection does not change the edge, but stake size and speed change total exposure. A lower stake per spin or a slower table reduces the amount put through the game, which reduces the expected loss in absolute terms. It does not make the game favourable; it makes the session smaller.
Progression Systems Redistribute Outcomes
A progression system is any rule that changes the stake based on previous results. The most common is doubling after a loss, sometimes called a martingale. The logic is that a win eventually recovers all previous losses plus one unit.
The logic is arithmetically true in a world with no table limits and no bankroll limit. In that world, a player with infinite funds and an operator with no maximum stake could eventually recover. The real world has both limits, and the progression collides with them.
As an example, suppose a player starts with a 1-unit bet on an even-money chance and doubles after each loss. After six consecutive losses, the next stake would be 64 units, and the total already lost would be 63 units. After ten consecutive losses, the next stake would be 1,024 units. The table maximum and the player's bankroll are both finite, and a losing streak long enough to exhaust either one produces a loss that no subsequent win can recover within the same progression.
The deeper point is that a progression does not change the expected value of any individual spin. Each spin still carries the same house edge. The progression changes the distribution of outcomes: most sessions end with a small win, and a small number of sessions end with a large loss. The large losses are exactly large enough to offset the small wins over time. The expected loss per unit wagered is unchanged.
No betting pattern changes the built-in edge of a negative-expectation game. Progression systems redistribute outcomes rather than remove the edge.
What the Table Rules Tell You Before You Bet
Before placing a chip, a player can read the table itself. The layout shows the payout for each bet type, and the wheel shows the pocket count. The wheel is the more important document, because the payout is meaningless without the pocket count behind it.
A table with a single zero is a European layout. A table with a single zero and a double zero is an American layout. Some tables add a rule that returns half of an even-money bet when the ball lands in zero, sometimes called la partage or en prison. That rule changes the expected loss on those even-money bets specifically, because it returns part of the stake on a losing outcome. The rule is printed on the table or in the game information panel, and it applies only to the bets it names.
The table minimum and maximum are also posted. These do not change the edge, but they set the range within which a progression can operate. A wide range allows longer progressions; a narrow range forces a stop sooner. Neither range makes the game favourable.
FAQ
Which roulette bet has the best odds?
On a single-zero wheel, all standard bets carry the same expected loss per unit, because the payout is short by the same amount relative to the probability on every bet. The bet type changes variance, not the long-run edge. The meaningful comparison is between wheel types: a single-zero wheel has a lower house edge than a double-zero wheel because it has one fewer losing pocket.
Can you improve roulette odds with a betting system?
A betting system changes the size and sequence of stakes, not the probability of any spin. Each spin still carries the same house edge, so the expected loss per unit wagered is unchanged. Progression systems redistribute outcomes, producing many small wins and occasional large losses, but the long-run arithmetic is the same as flat betting.
What does 35 to 1 mean in roulette?
It means a winning straight-up bet returns 35 units of profit for each unit staked, plus the original stake is returned. On a 37-pocket wheel, a fair payout for a 1-in-37 chance would be 36 to 1. The one-unit gap between the fair payout and the advertised payout is the house edge on that bet, and the same gap appears in every other payout on the layout.
Please play responsibly. If you or someone you know has a gambling problem, call or text the helpline printed in the footer below.
Play responsibly
Gambling should be treated as paid entertainment, never as a way to earn income or recover losses. 1-800-MY-RESET (1-800-697-3738)
18+ or 21+ depending on where you are; follow the minimum age that applies to you.
This article is general information about how these mechanics work. It is not legal advice and not a recommendation to gamble or to use any particular operator. Availability and legality differ by jurisdiction — check the rules that apply where you are.