Start at 0.01 lots. Lose, go to 0.02. Lose again, 0.04. By the tenth trade in the sequence the position is 5.12 lots, and the ten trades before it have already burned through 10.23 lots of losses. Nothing about the market changed across those ten trades. The only thing that grew was the size, and it grew by a factor of 512.
That is martingale. It comes from casino betting, where the doubling rule genuinely does recover the initial stake on the first win, provided the player has enough chips to survive the streak and the table has no maximum bet. Trading accounts have both limits: a balance, and a margin ceiling enforced by the broker.
The arithmetic nobody prints on the sales page
After n consecutive losses, the next position is 2 to the power of n times the base stake, and the total already lost is that number minus one. It compounds far faster than intuition suggests.
| Losses in a row | Size of the next trade | Already lost, in base units |
|---|---|---|
| 1 | 2x base | 1x |
| 5 | 32x base | 31x |
| 8 | 256x base | 255x |
| 10 | 1,024x base | 1,023x |
| 12 | 4,096x base | 4,095x |
Twelve losses in a row is not a freak event. Any trader taking several setups a day for a year will meet a streak of that length, and a trending market that refuses to mean revert produces them routinely. The system needs the trader to place a position four thousand times the starting size at exactly the moment their account is smallest and their nerve is thinnest.
Margin is the real ceiling
The sequence rarely gets to trade twelve, because margin requirements stop it first. Each new position locks up more of the account as required margin while the open losses eat the free equity. Required margin climbs, free equity falls, and the two curves meet. At that point the broker closes positions automatically, usually the largest one, which is the one carrying the entire recovery hope.
This is why martingale accounts do not bleed out. They stop in a single session. The equity curve is flat and pleasant for weeks, then vertical.
The dangerous part is that the strategy is most convincing right before it fails. A martingale account that has survived three months has simply not met its streak yet, and the position sizes it now runs are far larger than the ones it started with.
Why it feels like a working system
Martingale converts a low win rate into a high one. Almost every sequence ends in a win, because the sequence only ends when it wins. A trader reviewing their history sees a long column of green closes and concludes the entry logic is sound. The entry logic is usually irrelevant. A coin flip with a doubling rule produces the same green column.
What is hidden is the distribution of the loss. Standard risk statistics do not describe it well, because the average loss is tiny and the tail loss is the entire account. Reading drawdown on a martingale account tells you almost nothing until the day it tells you everything.
The behavioural version of the same trap is worse. Doubling after a loss is the mechanical form of revenge trading, and a trader who does it by hand escalates faster than any script would, because the decision to double is made while angry.
Grids, averaging down and recovery robots
Martingale rarely markets itself under that name. It appears as an expert advisor with a "recovery mode", a grid system that adds a position every 20 pips, or a discretionary habit of averaging down into a losing trade because the level "still looks good". The shared feature is that exposure increases while the position is losing, and the plan depends on price returning.
Grid systems are the most common disguise, which is why risk teams treat martingale and grid detection as one problem. Both look identical in the trade log: a cluster of same-direction entries at worsening prices, each larger than the last, closed together on the reversal.
What firms do about it
Any firm that puts real capital behind a trader has to detect this pattern, because an evaluation cannot distinguish a martingale trader from a disciplined one on results alone. The detection is straightforward once the data is structured: look for consecutive same-symbol entries in one direction with monotonically increasing volume while floating profit is negative, then compare the largest position in the account to the median position size. A ratio in the hundreds settles the question.
Firms that run a proper risk stack, whether built in house or on a platform like the Prop Firm CRM, flag the account on the first sequence rather than after the payout request. Explaining a disqualification is much easier when the rule fired on day two and the trader was told, and it is the fairest way to run the rule.
The alternative is dull and it works
Fixed fractional sizing does the opposite. Risk a constant small percentage of equity on each trade, so that size falls as the account falls and rises as it recovers. A losing streak shrinks the positions instead of inflating them, which is why the account can survive twelve losses and still be trading. The trade-off is honest: recovery is slow, and there is no mechanism that hides a bad edge behind a green column. Pair it with the sizing floor and daily stop described in the risk management rules, and a streak becomes an inconvenience rather than an ending.
"Martingale does not lose money slowly. It gives you a beautiful six months and takes all of it back in one afternoon."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- After ten straight losses the doubling rule demands a position 1,024 times the original size, funded by an account that has already lost 1,023 base units.
- Margin, not conviction, ends the sequence: required margin rises as free equity falls, and the broker closes the largest position first.
- Grid systems, recovery robots and averaging down are the same trade wearing different names, and they show the same signature in the trade log.
- Fixed fractional sizing shrinks positions during a losing run, which is the property that keeps an account alive long enough for an edge to matter.
Frequently Asked Questions
Does martingale ever work in trading?
It works right up to the losing streak that exceeds the account. The doubling rule needs unlimited capital and unlimited margin to guarantee recovery, and a retail trading account has neither. A long run of smooth gains followed by one account-ending loss is the expected shape of the strategy, not a malfunction of it.
Is grid trading the same as martingale?
They are close relatives. A grid adds positions at fixed price intervals rather than after each loss, and many grid systems also increase size at each level. Both build exposure against an open loss and both depend on price reversing before margin runs out, which is why risk teams usually treat them under the same rule.
Why do prop firms ban martingale if the trader paid for the evaluation?
Because a funded account is the firm's capital and the firm may hedge or copy it to a live venue. A doubling system produces an equity curve that looks stable for weeks, so it can pass an evaluation without demonstrating any risk control. Firms detect the size progression in trade data and disqualify it under prohibited strategy rules.