A red or black roulette bet covers 18 numbered pockets and pays even money. It is not a 50/50 wager because zero is neither color. On a single-zero wheel, the chosen color wins on 18 of 37 pockets. On a double-zero wheel, it wins on 18 of 38. A triple-zero wheel lowers the hit rate again to 18 of 39.
The payout stays 1 to 1 while the number of losing green pockets changes. That is the entire price difference.
The color layout does not contain a hidden imbalance
The numbers 1 through 36 are divided evenly:
- 18 are red;
- 18 are black;
- 0 is green;
- 00 and 000, where used, are also green.
Red and black therefore have identical probability on the same wheel. The sequence of colors around the wheel is not a simple alternation, but that physical order does not give one color more pockets.
| Wheel | Red wins | Black wins | Green losses | Probability of chosen color | Standard house edge |
|---|---|---|---|---|---|
| Single zero | 18 | 18 | 1 | (18/37=48.6486%) | 2.7027% |
| Double zero | 18 | 18 | 2 | (18/38=47.3684%) | 5.2632% |
| Triple zero | 18 | 18 | 3 | (18/39=46.1538%) | 7.6923% |
A winning 20-unit wager produces 20 units of profit and returns the original 20-unit stake. A losing spin removes the 20 units. “Even money” describes the payout, not the probability.
Expected value shows the cost in one line
For a 1-unit color bet, let:
- (W) be the number of winning color pockets;
- (L) be the number of losing pockets;
- (N=W+L) be the total pockets.
Because the profit and loss are both 1 unit:
[ EV=\frac{W}{N}(+1)+\frac{L}{N}(-1)=\frac{W-L}{N} ]
On a single-zero wheel:
[ EV=\frac{18-19}{37}=-\frac{1}{37}=-0.027027 ]
The house edge is therefore 2.7027%.
On a double-zero wheel:
[ EV=\frac{18-20}{38}=-\frac{2}{38}=-0.052632 ]
The house edge is 5.2632%. The same visible bet costs almost twice as much because of one additional green pocket.
This is why wheel selection matters more than choosing red or black. The main roulette house-edge guide compares the standard prices across the game, while European versus American roulette focuses on the practical wheel choice.
A worked session: result and theoretical cost are different
Suppose a player bets 15 units on black for 80 spins.
[ Total\ action=15\times80=1{,}200\text{ units} ]
Expected loss on single-zero roulette:
[ 1{,}200\times0.027027=32.43\text{ units} ]
Expected loss on double-zero roulette:
[ 1{,}200\times0.052632=63.16\text{ units} ]
Neither figure predicts the session result. Eighty spins are far too few for results to settle near expectation. The player can finish ahead on the more expensive wheel or lose heavily on the cheaper one. Expected value prices the repeated action; variance determines how widely an actual session can move around that price.
The expected-loss calculator is useful only after entering total action, not merely the amount brought to the table. Rebetting the same chips creates new wagering volume each spin.
Why color streaks are ordinary
A run of one color often feels like evidence that the opposite color is due. It is only a sequence.
On a single-zero wheel, the probability that the next eight spins are all red is:
[ P(8\ red)=\left(\frac{18}{37}\right)^8\approx0.314% ]
That is roughly 1 in 318 for a specific eight-spin block. “Eight of either color” is about twice that probability because the block could be all red or all black.
A rare-looking run becomes unsurprising when many overlapping blocks are observed across many tables and hours. The run does not alter the pocket count for the next spin. After seven reds, black still has 18 winning pockets on a single-zero wheel; red also has 18; zero remains the extra losing result for either color bet.
This is the central error behind the gambler’s fallacy and most color-progressions. Past independent results can describe what happened without changing what the wheel now offers.
Betting both colors does not create a hedge
A player who puts 10 units on red and 10 on black has 20 units in action.
- If a red number wins, the red bet earns 10 and the black bet loses 10: net 0.
- If a black number wins, the reverse happens: net 0.
- If a green pocket wins, both bets lose: net -20.
The position converts frequent zero-net spins into an occasional double loss. It does not remove the house edge.
On a single-zero wheel, expected value is:
[ EV=\frac{36}{37}(0)+\frac{1}{37}(-20)=-\frac{20}{37} ]
That equals a 2.7027% loss on the 20 units wagered. Covering both sides changes the result pattern, not the price per unit of action.
La Partage and En Prison require the exact table rule
Some single-zero tables protect even-money wagers when zero appears.
Under La Partage, a qualifying red or black wager normally loses half its stake on zero. The expected value of a 1-unit color bet becomes:
[ EV=\frac{18}{37}(+1)+\frac{18}{37}(-1)+\frac{1}{37}(-0.5)=-\frac{1}{74} ]
The house edge is 1.35135%.
En Prison can produce a similar or slightly different long-run value depending on the precise rule for the imprisoned stake, repeat zero, and subsequent settlement. Do not infer protection from a French-looking layout. Read the posted rule or help screen.
The lower edge applies only to qualifying even-money wagers. It does not automatically protect dozens, columns, streets, splits, or straight-up numbers.
The operational side of a simple bet
Color wagers are easy to describe but still require controlled settlement. The dealer must close betting before the ball falls, identify the winning number and color, clear losing outside bets, and pay winners at even money. Late-bet disputes and chips placed across layout boundaries are procedural issues, not mathematical ones.
Official roulette rules define red/black as a standard outside wager and identify zero pockets separately from both colors. The Nevada Gaming Control Board’s approved roulette rules provide the jurisdictional rule source; individual casinos may still apply different minimums, maximums, chip procedures, and approved variants.
From the casino’s viewpoint, color action also produces faster decisions than a layout crowded with many inside bets. Faster settlement can increase spins per hour. A low-looking wager repeated quickly may create more total exposure than a slower, more complex betting pattern.
A frequent win rate can still produce a losing session
Color bets win often enough to feel stable, but a win rate below 50% creates an awkward result distribution. With flat 1-unit bets over (n) spins, let (K) be the number of winning color results. Net profit is:
[ Profit=K-(n-K)=2K-n ]
The session finishes ahead only when (K>n/2). A tie in wins and losses merely returns the color bets to zero before any other costs.
Using the binomial model for independent spins, the approximate chance of finishing ahead is:
| Flat-bet session | Single-zero wheel | Double-zero wheel |
|---|---|---|
| 20 spins | 36.50% | 32.23% |
| 50 spins | 36.95% | 30.31% |
| 100 spins | 35.53% | 26.50% |
These percentages do not fall smoothly after every added spin because an even number of spins can finish exactly level. The broad direction is still clear: more repeated negative-expectation decisions make a profitable finish less likely.
The standard deviation of one 1-unit color result is close to 1 unit because each spin is almost always either +1 or -1. Over (n) flat bets, session standard deviation grows roughly with (\sqrt{n}), while expected loss grows directly with (n). This is why volatility dominates short sessions but the house edge becomes more visible as action accumulates.
A player can reduce dollar volatility by lowering the unit size. That does not change the percentage edge. Raising the stake after losses does the opposite: it concentrates more money into later spins without improving the probability of the chosen color.
The wheel history display does not add a third source of information
Electronic boards often show recent numbers, red/black counts, streaks, and percentages. Those displays can confirm that the last results were recorded, but they do not change the current pocket structure.
Suppose the board shows 14 black results in the last 20 spins. Two different statements must be separated:
- “Black occurred 14 times in the recorded sample” is a fact about the past.
- “Red is now more likely because the sample is unbalanced” is an unsupported prediction.
A balanced wheel does not correct short samples on a schedule. The long-run proportion approaches the underlying probability through many independent observations; the next spin does not owe the display a compensating color.
History boards can still have operational value. They help identify a disputed recorded outcome and may assist a casino in monitoring equipment. They are not a strategy signal for the player.
Decisions that actually improve the bet
There is no predictive choice between red and black. Practical improvement comes from controlling the terms around the wager:
- Prefer one green pocket to two or three.
- Prefer a genuine half-loss rule when available at acceptable limits.
- Compare total stake per spin, not just the base chip value.
- Avoid using a color run as a reason to raise the next wager.
- Treat a table maximum as a risk boundary, not a challenge to design a progression around.
The roulette odds calculator can verify pocket probabilities. The odd-or-even odds and high-or-low odds pages show the same 18-pocket structure with different labels.
Red and black are equally likely. The meaningful choice is not the color. It is which wheel, which zero rule, what stake, and how much total action you are prepared to place at a negative expectation.