The short answer: Martingale wins small amounts often and loses big amounts rarely. On Stake Dice the rare big losses cost more than all the small wins put together. On average you lose about 1% of everything you bet, exactly as you would betting the same amount flat. Martingale changes how a session feels. It does not change how it ends on average.
What is Martingale?
You bet 1 unit at a 2x payout, which on Stake Dice is a 49.5% win chance. Each time you lose, you double the bet. When you win, you go back to 1 unit. Each win gets back every loss in the streak, plus 1 unit of profit.
Why it feels like it works
Most losing streaks are short, so most sessions end a little ahead. With 1,023 units you can survive 10 losses in a row (1 + 2 + 4 and so on up to 512). The chance of losing 10 in a row is about 1 in 927, which sounds like never.
Why it does not
1 in 927 is the chance for one go. You are not having one go. In 1,000 bets at a 49.5% win chance there is about a 41% chance of hitting at least one 10-loss streak, and that one streak takes the whole 1,023 units.
| Losses in a row | Chance each time | Bankroll it takes (units) |
|---|---|---|
| 5 | 1 in 30 | 31 |
| 7 | 1 in 119 | 127 |
| 8 | 1 in 236 | 255 |
| 10 | 1 in 927 | 1,023 |
| 12 | 1 in 3,635 | 4,095 |
| 15 | 1 in 28,225 | 32,767 |
A bigger bankroll does not fix it. It makes the bust rarer and bigger by the same amount.
What the simulator shows
We ran this through the same simulator that is built into the DegenHelper script. Each session starts with a bankroll, bets 1 unit with Martingale at a 2x payout, and stops when it is up by the target or when the next bet no longer fits. Each row is 50,000 sessions.
| Bankroll | Stop when up by | Reached the target | Went bust | Average bet in total | Average cost |
|---|---|---|---|---|---|
| 127 units | 10 | 92.0% | 8.0% | 70 units | 0.7 units |
| 127 units | 100 | 43.3% | 56.7% | 489 units | 4.9 units |
| 1,023 units | 10 | 98.9% | 1.1% | 105 units | 1.1 units |
| 1,023 units | 100 | 89.9% | 10.1% | 991 units | 9.9 units |
Look at the last row. About 9 sessions in 10 win 100 units. The 1 in 10 that goes bust loses most of the 1,023. Put together, the average cost is still about 1% of what was bet, which is the house edge. The average cost column is exactly that: 1% of the total bet.
That is the whole trick of Martingale. It moves your losses out of sight, into one rare session, and makes every other session look like proof that it works.
If you use it anyway
- Set a stop-loss you would be fine hitting, not just the edge of your bankroll.
- Cap the bet size. A max bet stops the doubling before it eats everything.
- Bank wins as you go, so a bust cannot take what you already won.
- Run your own numbers before you bet anything.
The free DegenHelper script does all four on Stake: its Dice tool has a Martingale preset with a max bet, a stop-loss and a stop-when-up, a simulator that runs 500 practice sessions of your settings, and it saves a share of every win into your own vault.
To see what any win chance pays and what a bet costs on average, use the Stake Dice calculator.
FAQ
Does Martingale work on Stake Dice?
It wins often, but not on average. Every bet costs about 1% of its size on average, so the system costs about 1% of everything you bet, the same as betting flat.
What about reverse Martingale or other systems?
Same result. No pattern of bet sizes changes the odds of the next bet. Our guide to expected value shows why.
Is there a safe bankroll for Martingale?
No bankroll removes the risk. A bigger one makes a bust rarer and larger.
Why does it feel like it is working?
Because most sessions do end a little ahead. The loss sits in the rare session where a long losing streak arrives, and that session is big enough to cancel the rest.
18+ only. Gambling can be addictive. If it has stopped being fun, Play responsibly lists free, confidential help.