Roulette expected loss per hour is not a prediction of what your next session will lose. It is a pricing estimate: how much negative expectation is attached to the total amount you are likely to wager during an hour. That makes four inputs matter immediately—wheel rules, average stake per spin, number of spins, and whether you spread chips across several bets at once.
Start with hourly action, not the cash you bought in with
A buy-in tells you how much money entered your rack. It does not tell you how much money entered action. If you buy in for $200, keep recycling chips, and make sixty $10 wagers, you have created $600 of action even though your starting cash was much smaller.
That distinction is the whole reason an hourly estimate is useful. Roulette does not charge a time-based fee. The mathematical cost appears when chips are exposed to a payout schedule that is shorter than the true odds.
For a standard single-zero wheel, the usual house edge is 1/37, about 2.70%. For a standard double-zero wheel, it is 2/38, about 5.26% on ordinary wagers. The current Wizard of Odds roulette basics lists those standard figures, while regulated rules such as the Nevada Live Roulette rules show the familiar 35:1, 17:1, 11:1, 8:1, 5:1, 2:1 and 1:1 payout structure.
Build the estimate from four inputs
The basic relationship is:
$$Hourly\ Expected\ Loss = Average\ Stake\ Per\ Spin \times Spins\ Per\ Hour \times House\ Edge$$
The arithmetic is easy; defining the stake correctly is where players often go wrong.
Suppose a player says, “I bet $10 a spin,” but the actual layout contains $10 on black, $5 on the first dozen and four $1 straight-up numbers. The real average stake is $19. If the table produces 55 spins in an hour, hourly action is $1,045—not $550.
At single-zero pricing, $1,045 of ordinary action has baseline expected loss of about $28.24. At double-zero pricing, the same action has baseline expected loss of about $55.00. The player can finish up $300 or down $500 in that hour; expected loss is the long-run price attached to the volume, not a cap on what variance can do.
Compare three very different one-hour sessions
| Session | Average stake | Spins | Total action | Applicable edge | Baseline expected loss |
|---|---|---|---|---|---|
| Single-zero live table | $10 | 45 | $450 | 2.70% | about $12.16 |
| Double-zero live table | $10 | 45 | $450 | 5.26% | about $23.68 |
| Single-zero terminal | $10 | 120 | $1,200 | 2.70% | about $32.43 |
The terminal example is the important one. A better wheel can still produce a larger expected dollar cost per hour if it produces far more decisions. This is why roulette spin speed and total action belongs beside roulette house edge, not underneath it as a minor detail.
Mixed bets require a weighted price when the edges differ
On a normal single-zero wheel without special rules, ordinary straight-up, split, street, corner, six-line, dozen and even-money wagers all carry the same 2.70% edge, so adding them together is straightforward. On a standard double-zero wheel, ordinary wagers are generally 5.26%, while the five-number 0-00-1-2-3 bet is worse.
French-style rules introduce another wrinkle. If La Partage applies to an eligible even-money wager, the effective edge on that wager can fall to about 1.35%, while inside bets remain at the normal single-zero price. A player mixing $20 of eligible even-money action with $10 of inside action should therefore not apply one percentage blindly to the whole $30.
A more precise estimate is:
$$Expected\ Loss = \sum (Action_i \times Edge_i)$$
That weighted approach is also the right way to evaluate electronic roulette with nonstandard payouts. First identify the actual wager and rule set; then price the action.
Pace can change without the player noticing
Live roulette contains natural friction: chips must be placed, the dealer closes betting, the ball settles, losing bets are cleared, winners are paid, and the layout is reopened. Stadium and electronic formats can compress some of that cycle and make repeat wagers almost effortless.
The result is behavioral as much as mechanical. A player who would reconsider each $20 live-table bet may press “repeat” rapidly on a terminal. The nominal unit stays unchanged while hourly action climbs.
This is why “I only play small stakes” is incomplete. A small stake multiplied by many decisions can create more action than a larger stake at a slow table.
Expected loss and bankroll risk answer different questions
Expected loss is an average-cost measure. Bankroll risk is about the distribution around that average. Roulette has high short-term variance, especially when a player uses inside numbers or concentrates action on a small number of pockets.
Two players can have the same expected loss but very different session experiences. One may flat bet $20 on red; another may place $20 on a single number. If both use the same wheel and create the same total action, the underlying house-price percentage is similar, but the frequency and size of wins are radically different.
Use roulette bankroll risk for the second question. Expected loss tells you the mathematical price of volume; bankroll analysis tells you how violently the path can move before the long run becomes visible.
Casino theoretical win is the mirror image of this calculation
From the operator side, average bet, decisions per hour and game edge feed theoretical win. A roulette player who creates $2,000 of action on a 2.70% game carries roughly $54 of theoretical loss before any comp reinvestment assumptions are applied.
That is why a casino rating can look very different from the player’s actual result. A player may win $500 but still generate positive theoretical value for the house. Another may lose $500 quickly after very little action and generate much less theo than the raw loss suggests.
For a deeper operator view, see roulette comp value. The principle is the same calculation seen from opposite sides of the table.
A practical worksheet for estimating one hour
Before the session, write down five numbers:
- wheel type and any special zero rule;
- expected average total stake per spin;
- a realistic spin count for that format;
- the edge attached to each wager type if they differ;
- the maximum session length you actually intend to play.
Then calculate total action and expected loss. If you change stake size or move to a faster game, recalculate instead of assuming the original estimate still applies.
For example, $15 average action for 50 spins creates $750 of volume. On ordinary single-zero roulette, expected loss is about $20.27. On ordinary double-zero roulette, it is about $39.47. The difference is not a forecasting trick; it is the cost difference built into the wheel.
The expected loss calculator can automate the arithmetic, but the quality of the answer still depends on honest inputs.
Read the number as a price, not a prophecy
A player can lose far more than expected loss in an hour, win despite negative expectation, or finish almost exactly on the estimate. Short sessions are noisy. The estimate becomes meaningful as a way to compare choices: single zero versus double zero, slow live table versus fast terminal, $10 average action versus $25, thirty minutes versus three hours.
That is the useful question roulette expected loss per hour answers: How expensive is the amount of action I am choosing to buy under these rules? Once framed that way, wheel quality and playing speed become part of the same decision instead of separate trivia.