Skip to the page
Expected Value

Binomial Distribution Expected Value for Roulette Bets

Expected wins equal n times p. Worked on 38 straight roulette bets: the win count, how often a session sees no win, and the net under New Jersey payout rules.

By TrumpCasino Guides6 min read

A polished wooden roulette wheel with red and black numbered pockets and a brass centre spindle in a casino.

Photo: Pavel Danilyuk on Pexels

Jump to a section
  1. 01What does the binomial expected value formula say?
  2. 02Is a roulette bet a binomial trial?
  3. 03How to do expected value for a session of straight bets
  4. 04How far does a session stray from n × p?
  5. 05Why even-money bets break the clean binomial
  6. 06What the expected value of a binomial distribution does not tell you
  7. 07Guide questions
  8. What is the expected value of a binomial distribution?
  9. Is a single-number roulette bet a binomial trial?
  10. Why can a session end with no wins when the expected count is one?
  11. Does the formula give my net result as well as my wins?
  12. 08Sources and further reading

The expected value of a binomial distribution is n × p: the number of trials multiplied by the probability of success on each trial. Place 38 straight-up bets on a double zero roulette wheel, where each bet has a 1 in 38 chance, and the expected number of wins is exactly 1. What that single expected win is worth, and how far a real session drifts from it, depends on the payout attached to the bet.

What does the binomial expected value formula say?

The binomial distribution expected value is the average number of successes across many repeats of the same fixed run of trials. The model counts successes in N trials where each trial has the same success probability p, and the mean of that count is simply n × p.

Three conditions have to hold before the formula applies to a wager:

  • A fixed number of trials, decided before you start (a session of a set number of spins).
  • Two outcomes per trial: the bet wins or it does not.
  • A probability that does not move, since the model assumes p is fixed for every trial.

The binomial counts how many wins land inside a fixed run. Its sibling, the geometric random variable expected value, counts how long you wait for the first one. Both lead back to the same p.

Is a roulette bet a binomial trial?

A single-number roulette bet is a clean binomial trial: one winning compartment, a fixed chance, and a win or a loss on every spin. The wheel itself sets p. New Jersey rules require that a double zero wheel has 38 compartments, the 0, the 00 and the numbers 1 to 36, so a straight bet wins with p = 1/38. A single zero wheel has 37 compartments, so p = 1/37 there.

The payout is set by regulation too. As of November 2026, the New Jersey code says casinos may not pay winning roulette wagers below the listed odds, and the listed odds for a straight bet are 35 to 1. With p and the payout both fixed in rule text, every number in the sums below follows from them.

How to do expected value for a session of straight bets

To turn the binomial mean into a session result, multiply the expected wins by what a win pays, then subtract the stakes lost on every other spin. One unit on a straight bet that wins returns 35 units of profit plus the stake; every losing spin costs the 1 unit staked.

WheelSpins (n)p per spinExpected wins (n × p)Units stakedProfit from one winUnits lost on the other spinsExpected net
Double zero381/381383537−2
Single zero371/371373536−1

The pattern is the point. The expected win count is the same on both wheels, but the extra 00 compartment adds a spin you are expected to lose, and that is where the double zero wheel's extra cost lives.

How far does a session stray from n × p?

The expected value of a binomial distribution tells you the centre, not the session you will actually play. On a double zero wheel, a run of 38 straight bets has three broad outcomes, worked from p = 1/38 with the binomial probabilities:

Wins in 38 spinsChance of this outcomeNet result at 1 unit per spin
0about 36%−38
1about 37%−2
2 or moreabout 26%+34 or better

So the single most likely result is the expected one, yet more than a third of these sessions end without a win, and roughly a quarter end ahead. That spread (the binomial variance, which gets its own guide) is why a stake plan matters more than the average. A bankroll management calculator sizes a session so the zero-win branch of this table does not end it.

Why even-money bets break the clean binomial

An even-money roulette bet still has a binomial count of wins, but its net result per spin has three outcomes, not two. Take the 1 to 18 bet on a double zero wheel: 18 winning numbers, so p = 18/38, and over 38 spins the expected wins are 18.

The losses are where it changes. In New Jersey, when the ball lands on 0 or 00, the player loses either half or the entire wager, at the casino's option, on red, black, odd, even, 1 to 18 and 19 to 36. A spin can win 1 unit, lose 1 unit or lose half a unit.

Rule on 0 and 00Expected wins in 38 spinsUnits lost on 19 to 36Units lost on 0 and 00Expected net
Entire wager lost18182−2
Half wager lost18181−1

The win count is pure binomial; the net is not, because a third outcome sits on two compartments. Raising stakes after losses, as in a roulette martingale strategy, changes the shape of a session's results but leaves the expected net per unit staked where these tables put it.

What the expected value of a binomial distribution does not tell you

The formula assumes the same p on every trial and a fixed number of trials. It says nothing about when wins arrive, how long a losing run lasts or whether a session ends early. Use n × p for the centre, the payout for the net, and the spread for the bankroll. More worked sums sit in the expected value guides on this site.

Guide questions

What is the expected value of a binomial distribution?

It is n times p: the number of trials multiplied by the chance of success on each one. It gives the average count of successes over many repeats of the same run, not the count you will see in any single run.

Is a single-number roulette bet a binomial trial?

Yes. Each spin is a separate trial with one winning compartment and a probability that does not change from spin to spin, so the number of wins in a run of spins follows the binomial model.

Why can a session end with no wins when the expected count is one?

The expected value is an average across many sessions. A single run of straight bets is spread around that average, and a run with no wins at all is one of the most common outcomes.

Does the formula give my net result as well as my wins?

Only after you attach the payout. The binomial mean counts wins; the net comes from multiplying wins by the payout and subtracting the stakes lost on every other spin.

Sources and further reading

Choosing where to play? See the casinos we cover and our take on each.

Where to play