Martingale System in Roulette: How It Works and Why It Fails
The martingale is a betting progression that doubles the stake after each loss. This guide shows the doubling table, the chance of a long losing run and the expected loss per sequence.
You may have landed here after reading about horse tack or a branch of probability theory. In gambling, a martingale is a betting progression: you double your stake after every loss, so that one win recovers all earlier losses and leaves a profit equal to your first bet. On roulette it is almost always played on an even-money bet such as red or black.
This guide shows what a martingale does to your results, with the arithmetic. The short version: it changes how wins and losses are spread out, not the house edge. For adults 18+ only, and nothing here is a promise of profit.
What Is a Martingale in Roulette?
If you ask "what is a martingale", the roulette answer is one rule. Bet one unit on red. If it loses, bet two. If that loses, bet four. When red finally hits, you win back everything lost in the run, plus one unit.
Even-money bets pay 1 to 1. A winning bet of 8 units returns 16, which covers the 7 units lost before it (1 + 2 + 4) plus the 8 just staked, with 1 unit of profit left over. That is the whole appeal: most sequences end with a small win.
The catch is what happens on the sequences that do not.
The Martingale Doubling Table
Take a base stake of 10 and a table maximum of 500. These are example numbers; real limits vary by table.
| Loss in a row | Next stake | Total staked so far | Net result if this bet wins |
|---|---|---|---|
| 0 | 10 | 10 | +10 |
| 1 | 20 | 30 | +10 |
| 2 | 40 | 70 | +10 |
| 3 | 80 | 150 | +10 |
| 4 | 160 | 310 | +10 |
| 5 | 320 | 630 | +10 |
| 6 | 640 | 1,270 | +10 |
The stake after n losses is 10 × 2^n, and the total staked after n+1 bets is 10 × (2^(n+1) − 1). Every win nets +10, no matter how long the run was.
Now look at the last row. The next stake would be 640, which is above the 500 limit. After six straight losses you have lost 630 and cannot place the bet that would recover it. The progression is cut off at the moment it matters most.
A smaller limit makes it worse. At a 200 maximum, the fifth bet (160) is the last one allowed, and the run ends after 310 lost.
The Odds of Losing N Spins in a Row
On a European wheel there are 37 pockets. Red wins on 18, so the chance of losing a red bet is 19/37, or 51.35%. Spins are independent, so the chance of N losses in a row is (19/37)^N.
| Losses in a row (N) | Probability | Roughly 1 in |
|---|---|---|
| 1 | 51.35% | 2 |
| 3 | 13.54% | 7 |
| 5 | 3.57% | 28 |
| 6 | 1.83% | 55 |
| 7 | 0.94% | 106 |
| 8 | 0.48% | 207 |
| 10 | 0.13% | 784 |
Those numbers are per sequence, not per session. If you start a fresh sequence after each win, you run many sequences, and a 1.83% chance repeated enough times becomes likely. Fifty sequences give 1 − (1 − 0.0183)^50 = 60% that at least one of them hits six losses, by my calculation.
Expected Loss per Sequence
Use the table above: base 10, and a sequence that ends at the first win or after six straight losses (630 lost).
- Chance of reaching a win within six bets: 1 − 0.018336 = 98.17%, net +10.
- Chance of six losses: 1.83%, net −630.
Expected result = 0.981664 × 10 + 0.018336 × (−630) = 9.817 − 11.552 = −1.74 per sequence.
You can check it a second way. The expected total staked in a sequence is 10 × (1 + 38/37 + (38/37)² + … up to six terms) ≈ 64.20. Multiply by the house edge, 1/37 = 2.70%: 64.20 ÷ 37 = 1.74. Same answer, as it must be.
So the martingale does not remove the edge. It collects it in an unusual shape: about 98 sequences in 100 win 10, and roughly 2 in 100 lose 630. The average is still negative, and it is exactly the edge times the money you put at risk.
On an American double-zero wheel the edge is 2/38 = 5.26%, the loss chance per bet is 20/38, and the same table gives a six-loss chance of 2.13% and an expected loss of about 3.60 per sequence.
Why the Martingale Feels Like It Works
A long run of wins is the expected experience. The method wins small amounts often, so a short trial can look flawless. A losing run of six or seven is rare enough that many players never see it before they quit, and then it arrives as a single large loss.
There is also the belief that a long losing run makes a win "due". It does not. After five losses on red, the next spin is still 18/37 for red. The wheel keeps no memory. Streak counts, including the live ones on this site, describe what already happened. They do not move the next probability.
Other Limits That Break the Progression
The table maximum is the hard wall, but not the only one:
- Bankroll. To survive six losses at base 10 you need 630, and to bet on the seventh you need 1,270. Needing 1,270 to win 10 is the shape of the risk.
- Time. Long sessions mean more sequences, which means more chances to meet the long run.
- Stake size at the end. A final bet of 320 or 640 is a different decision from a bet of 10, even though the system treats them as the same.
Variants such as the Labouchère or D'Alembert move the risk around differently. None of them changes the edge on a single spin, and the expected loss still equals the edge times total staked.
If You Play Anyway: Set Limits First
Treat it as paid entertainment, not an income plan. Decide the loss you can afford before you start, set a stop rule such as "I stop when I have lost my budget or after 45 minutes, whichever comes first", and never raise a stake to chase a loss you cannot cover. If gambling stops being fun, take a break and use your local self-exclusion or support services.
Roulette Streaks shows live streak data and Telegram alerts from real tables. It is descriptive information, and it does not improve your odds on any bet, martingale or otherwise.
FAQ
Does the martingale work in roulette?
It wins a small amount in most sequences and loses a large amount in a few. The expected result is still negative, at the house edge times the amount staked: 2.70% on a European wheel, 5.26% on an American one.
What is the chance of losing 6 times in a row on red?
On a European wheel, (19/37)^6 = 1.83%, about 1 in 55. On an American wheel, (20/38)^6 = 2.13%.
Can a bigger bankroll make the martingale safe?
A bigger bankroll delays the point where you hit a limit, but table maximums still cap the progression, and more bets staked means a larger expected loss in total.
Is the martingale the same as the gambler's fallacy?
They are linked. The martingale assumes a win is coming soon after a loss streak, and the gambler's fallacy is the same belief. Each spin is independent, so the odds stay the same.