Roulette Expected Value (EV) Explained
What this guide covers
Expected value (EV) is the single calculation that ties probability, payouts, and house edge together into one number describing what a bet is "worth" on average. This guide covers how to calculate EV for any roulette bet, why virtually every bet on the table has the same EV, and how to use that understanding practically. For the underlying probability figures this guide builds on, see roulette probability explained; for how EV relates to the house's side of the math, see roulette house edge explained.
What expected value means
Expected value is the average outcome of a bet if you could somehow repeat it an infinite number of times, calculated by multiplying each possible outcome by its probability and summing the results. It's not a prediction of what will happen on any single spin — it's a long-run average that smooths out every possible win and loss into one representative number, expressed either as a dollar amount or as a percentage of the stake.
The EV formula
EV = (probability of winning × amount won) − (probability of losing × amount lost). For a $1 straight-up bet on European roulette: your probability of winning is 1/37, paying $35 in profit; your probability of losing is 36/37, costing your $1 stake. The full calculation: (1/37 × $35) − (36/37 × $1) = $0.9459 − $0.9730 = −$0.027. That negative $0.027 is your expected loss per dollar wagered — exactly the 2.70% house edge expressed in dollar terms.
Every standard bet has (almost) the same negative EV
This is the concept that most surprises new players: whether you bet straight up, on a split, on a dozen, or on red/black, the EV per dollar wagered on a European wheel works out to the same −2.70% (or −$0.027 per $1 bet) in every case. Higher-payout bets have lower win probability; lower-payout bets have higher win probability — and the two always cancel out to the identical negative EV, because they're all calibrated against the same single zero pocket. The one notable exception is the American wheel's five-number bet, whose awkward odds give it a worse EV (roughly −7.89%) than every other bet at the same table — see roulette betting options explained for the full breakdown.
Worked example: a full betting session
Suppose you place 100 spins of $10 each on European roulette's red bet, for $1,000 total wagered. Your expected loss, using the −2.70% EV, is $1,000 × 0.027 = $27. Your actual result after 100 real spins will almost certainly differ from exactly −$27 due to variance — you might finish up $150 or down $200 — but −$27 remains the mathematically "expected" center of that distribution, and it's the number that becomes increasingly accurate as you extend the session to thousands of spins instead of 100.
Why EV doesn't predict your session
EV describes an average across an enormous number of repetitions, not a forecast for your specific session. A single evening of roulette, even a long one, is subject to genuine variance that can push your actual result well above or below the theoretical EV. This is exactly why treating EV as "what should happen tonight" leads to confusion — it's a planning tool for understanding long-run cost, not a short-term prediction. See roulette probability explained for more on how variance and probability interact.
EV and bet sizing
Because EV scales directly with stake size, doubling your average bet doubles your expected loss over any given number of spins, while halving it halves the expected loss. This straightforward relationship is the mathematical backbone of bankroll management — your expected cost of playing is entirely predictable once you know your average bet size, the wheel's house edge, and roughly how many spins you plan to play. See roulette bankroll management guide for how to apply this when planning a session budget.
EV across wheel types
| Wheel type | House edge | EV per $10 bet |
|---|---|---|
| European roulette | 2.70% | −$0.27 |
| American roulette | 5.26% | −$0.53 |
| French roulette (la partage, even-money bets) | ~1.35% | −$0.135 |
This table makes concrete something that's easy to state abstractly: choosing European over American roulette for a session of, say, 100 average-sized bets is worth roughly double the difference in expected cost — a meaningfully larger number than most players intuitively expect from "just" a different wheel.
Why no betting system produces positive EV
Martingale, Fibonacci, and every other progression system redistribute how much you bet across a sequence of spins, but they don't change the probability or payout of any individual bet — which means they don't change that bet's EV either. Summing negative-EV bets, regardless of how their sizes are sequenced, always produces a negative total EV. A betting system can change your variance (how spread out your possible outcomes are) without ever making the underlying average outcome positive. See flat betting vs progressive betting and roulette strategy guide for more on this distinction.
EV thinking versus emotional decision-making
Understanding EV gives you a stable, unemotional reference point during a session — a losing streak doesn't mean the game has become less fair, and a winning streak doesn't mean you've found an edge; both are simply variance around the same fixed EV. Players who understand this tend to make steadier bet-sizing decisions than players chasing losses or increasing bets after a win, both of which are emotional responses that EV math doesn't support or require.
Using EV to compare games, not just bets
EV comparisons are also useful across entire games, not just individual bets. Blackjack played with correct basic strategy typically carries an EV much closer to zero (often under −1%) than roulette's −2.70% to −5.26%, which is one reason serious value-focused players often split time between the two. This doesn't mean roulette is a poor choice — it means EV is one more input, alongside pace and personal enjoyment, worth weighing when deciding how to allocate a gambling budget. See blackjack vs roulette for the full comparison.
Frequently asked questions
What does a negative EV actually mean in practice? It means that, averaged over a very large number of repetitions, you're mathematically expected to lose that percentage of every dollar wagered — it doesn't predict what happens in any individual session.
Do all roulette bets have the same EV? Almost all bets on a given wheel type share the same EV, since higher payouts and lower win probabilities offset each other identically; the American five-number bet is the main exception, with a notably worse EV than every other bet at the table.
Can a betting system make EV positive? No — betting systems change how stakes are sized and sequenced across a session, but since they don't change the probability or payout of any individual spin, the underlying EV per dollar wagered stays the same.
How is EV different from house edge? House edge is EV expressed as a percentage of the amount wagered; EV can also be expressed as an absolute dollar figure for a specific bet size, making it more directly useful for session planning.
Why did I win money even though EV is negative? Negative EV describes the long-run average outcome, not a guarantee for any single session — variance means individual sessions frequently finish above the theoretical average, just as they frequently finish below it.
Does EV change between European and American roulette? Yes — American roulette's extra zero pocket roughly doubles the negative EV per dollar wagered compared to European roulette, making wheel choice one of the most impactful decisions a player can make.
