Roulette expected value (EV) is the average profit or loss attached to a wager if the same pricing situation could be repeated a very large number of times. It is not a prediction that your next spin will lose 2.70%, and it is not a promise that a one-hour session will finish close to the average.
EV is better understood as a price tag. It combines probability and payout into one number so that bets with very different hit rates can be compared on the same scale.
Expected value is a price, not a forecast
Suppose a player stakes $10 on red on a European single-zero wheel. The wager has only two financial outcomes under ordinary rules: win $10 of profit or lose the $10 stake. Red occupies 18 of the 37 pockets; the other 18 black pockets plus zero lose.
The expected value is:
$$EV=\left(\frac{18}{37}\times 10\right)-\left(\frac{19}{37}\times 10\right)$$
$$EV=-\frac{10}{37}\approx -$0.2703$$
That does not mean the casino takes 27 cents from the wager in real time. The actual spin pays +$10 or -$10. The -$0.2703 is the long-run average value of putting $10 into that wager once.
The Wizard of Odds roulette basics lists the standard probabilities and payouts used in these calculations. Official rules such as the Nevada roulette rules of play and Massachusetts roulette rules show why EV must be tied to the actual wheel and settlement rule rather than to a generic game name.
Build a straight-up EV from the payout table
A one-unit straight-up number on European roulette wins on 1 of 37 pockets and pays 35 to 1. It loses on the other 36 pockets.
$$EV=\left(\frac{1}{37}\times 35\right)-\left(\frac{36}{37}\times 1\right)$$
$$EV=-\frac{1}{37}\approx -0.027027\text{ units}$$
Expressed as a percentage of the one-unit stake, that is about -2.70%.
The important part is the payout term. If the game paid true odds, the winning profit would need to compensate fully for all losing outcomes. Standard roulette pays 35 to 1 even though a single number on a 37-pocket wheel has 36 losing pockets. The missing unit is the casino’s price.
This is the cleanest bridge between roulette payouts and roulette house edge: expected value is the player-side result in units or money; house edge expresses the same disadvantage as a percentage of the stake.
Why red and a single number can share the same percentage EV
Red and straight-up roulette do not feel remotely similar. Red wins often and pays 1 to 1. A single number wins rarely and pays 35 to 1. Yet on a standard single-zero wheel they produce the same percentage expected value.
| Bet | Win probability | Net win | Loss probability | EV per 1 unit |
|---|---|---|---|---|
| Red | 18/37 | +1 | 19/37 | -1/37 |
| Straight-up | 1/37 | +35 | 36/37 | -1/37 |
This is why hit rate alone cannot tell you whether a wager is cheap or expensive. The payout must be evaluated at the same time.
It is also why roulette variance deserves its own concept. Two bets can have the same EV and very different distributions of outcomes. EV tells you the center of the long-run distribution; variance tells you how widely individual sessions can swing around that center.
Move from one wager to the expected cost of a session
Once the EV percentage is known, session pricing is straightforward.
If a player wagers $20 per spin for 60 spins on a standard European wheel, total action is:
$$20\times 60=$1{,}200$$
At a 2.70% standard edge, the approximate expected loss is:
$$$1{,}200\times 0.027027\approx $32.43$$
If the same $1,200 of action is placed on ordinary double-zero roulette at about 5.26%, expected loss is roughly:
$$$1{,}200\times 0.052632\approx $63.16$$
Those values describe averages over repetition. A real player can win $300, lose the whole bankroll, or finish almost even. EV does not become false because a particular session lands far from the mean.
The expected loss calculator is designed for this translation from percentage to money.
Rule changes alter EV only when they alter probability or settlement
A betting system that changes the next stake after a win or loss does not improve the expected value of the underlying roulette wager. A wheel or rule change can.
Moving from double zero to single zero removes one losing green pocket while keeping standard payouts. A French La Partage rule can return half of an even-money stake when zero lands, changing the settlement distribution and reducing the effective disadvantage on those bets.
The key diagnostic question is therefore:
What changed in the probability or the payout?
If nothing changed except the order of your bets, the sign of EV did not improve. This is the core reason systems described in why roulette systems fail cannot manufacture an edge through staking patterns alone.
Expected value and RTP are two views of the same price
Roulette pages often use both EV and return to player (RTP), which can sound like different concepts.
For a one-unit wager with an expected loss of 0.027027 units:
$$Player\ EV=-0.027027$$
$$House\ Edge=2.7027%$$
$$RTP=100%-2.7027%=97.2973%$$
RTP is the expected amount returned, including the original stake where appropriate in the accounting convention. House edge is the expected fraction retained by the house. EV can be expressed in units, dollars, or as a percentage.
For more on that relationship, use roulette RTP rather than treating RTP as a promise that every $100 session returns $97.30.
Why actual results can stay far from EV for a long time
Roulette is noisy. A player can hit a straight-up number twice in ten spins and finish with a huge profit despite negative EV. Another player can lose eight consecutive red bets and finish far below the average despite using a low-volatility wager.
Expected value becomes useful through repetition, but repetition does not force results to march smoothly toward the average. It increases the amount of data and total action. The distribution can remain volatile even while the average cost becomes increasingly meaningful to the casino across many players and tables.
This distinction prevents two common errors. A winning session does not prove a negative-EV wager became positive. A losing session does not prove the wheel was unfair. Both are possible outcomes inside a variable game.
If you want to explore that spread rather than just the average, use the variance simulator.
How casinos use EV without knowing the next number
A casino does not need to predict the next roulette result to price the game. It needs a known paytable, a controlled wheel, reliable procedures, and enough wagering volume for theoretical expectations to be useful at portfolio scale.
Suppose a table handles $50,000 of standard single-zero roulette action. At 2.70%, the theoretical win attached to that action is roughly $1,350. Actual table win for the shift can be much higher or lower. Over larger samples, theoretical figures help management compare game mix, player ratings, staffing, and performance without pretending that one shift should equal the mathematical average.
That is why EV is central to casino operations but still poor as a next-spin forecast. It is a pricing and planning metric.
A practical EV checklist for any roulette claim
When somebody claims a roulette bet or system is superior, run the claim through four steps:
- Define every possible outcome. Include zero and double zero where applicable.
- Assign the exact probability to each outcome. Use the actual wheel, not an imagined 36-number wheel.
- Apply the real payout or loss rule. Include half-back, prison, commissions, bonuses, or special baskets if the product has them.
- Multiply and sum. The resulting weighted average is the expected value.
Then separate EV from variance. A system may change how large the swings are. A package may change how many pockets are covered. A progression may change average stake. None of those facts alone proves an improved expected value.
The useful question is “What is each dollar of action worth on average?”
That framing is much more powerful than asking whether a bet “wins often” or whether a system “worked last night.” EV prices the wager on a common scale.
On standard roulette, the wheel type usually matters more than whether you choose red, a dozen, or one number. Bet selection then changes the ride: hit frequency, payout size, and variance. Session length and bet size determine how much money is exposed to that price.
Continue with roulette house edge for the percentage view, roulette odds for pocket probabilities, and roulette variance for session spread. The roulette odds calculator is useful when you want to construct EV from a specific set of covered pockets rather than relying on a slogan about the “best” wager.