The arithmetic
Each position is the previous one multiplied by a constant after a loss:
position at step n = first position × multiplier^(n − 1)
capital to survive n losses = first × (multiplier^n − 1) ÷ (multiplier − 1)
Doubling from a single unit, the sequence runs 1, 2, 4, 8, 16, 32, 64, 128, 256, 512 — and surviving all ten requires 1,023 units. Not ten times the first position. A thousand times.
That is the whole reason this page exists. Nobody plans to risk a thousand units. They plan to risk one, and the sequence quietly commits them to the rest.
| Losses in a row | Capital needed (doubling) | Final position |
|---|---|---|
| 5 | 31× | 16× |
| 8 | 255× | 128× |
| 10 | 1,023× | 512× |
| 12 | 4,095× | 2,048× |
| 15 | 32,767× | 16,384× |
Why it feels like it works
Because for a long time it does. Every completed sequence recovers all previous losses and adds one unit of profit, so the equity curve is a clean staircase — dozens of small green steps, then dozens more. It looks like an edge and reads like consistency.
What is actually happening is that the losses are being deferred rather than avoided. They accumulate off-screen, invisible until the streak arrives that the account cannot fund. Then the whole staircase is repaid at once, from the largest position in the sequence.
This is the honest description: martingale converts many small wins and one enormous loss into something that resembles a strategy. It does not change expectancy at all — a negative edge stays negative, and the doubling only decides the shape of the arrival.
The streak is not as unlikely as it feels
The intuition that ten losses in a row will not happen is where the system does its real damage.
At a 50% win rate, the probability of ten consecutive losses on any given attempt is about one in a thousand. Over a few thousand trades — a normal year for an active trader — it is close to certain. At a 60% win rate it is roughly one in ten thousand per attempt, which sounds safe until you notice that a few years of daily trading is exactly that many trades.
And the streak does not need to be improbable to matter. It only needs to happen once, because the position that ends the account is the largest one you have ever taken, placed at the worst moment you have ever had.
Where it shows up without the name
Martingale is rarely chosen deliberately. It arrives as behaviour:
- adding to a loser to "improve the average";
- doubling size after a losing day to get back to flat;
- widening a stop and increasing size because the level "still holds";
- grid systems, which are martingale with a schedule.
These feel like separate decisions in the moment. In the account they are one sequence, and the recovery arithmetic applies to what is left afterwards.