What Is Martingale strategy?
A martingale strategy increases the size of a position after each loss, so that a single win recovers all earlier losses plus a small profit. The name comes from an 18th-century betting system: double the stake after every lost coin flip, and the first win leaves you one unit ahead. In trading, “martingale” or “martingale-like” describes any position sizing rule that adds to a losing position in growing amounts, usually inside a mean reversion strategy that expects the price to bounce back.
A typical trading version places a ladder of buy orders below the current price, each larger than the one before (for example 1.5 to 2 times), and a take profit slightly above the average entry price. Each new fill lowers the average entry, so a smaller rebound closes the whole position in profit. Crypto bots, grid trading systems and some market-making vaults use this pattern. Averaging down is the same idea without the growing size.
Why it looks good and why it fails
Most price moves partially reverse, so most ladders close in profit. The result is a very high win rate, a smooth equity curve and a high Sharpe ratio, until a move keeps going. Position size grows geometrically: with doubling, after n losses the total committed is 2ⁿ − 1 units and the next order is 2ⁿ units. A trader who starts with 1% of capital has committed 63% after six losing steps and cannot place the seventh order of 64%. At that point the strategy holds its largest position at its worst price, and with leverage it faces liquidation. One such event can erase years of small gains.
The mathematics is unforgiving. In a fair game, no staking system changes the expected profit when capital and time are finite; this follows from Doob’s optional stopping theorem. A martingale only reshapes the distribution: many small wins, a rare ruinous loss, and a risk of ruin that approaches certainty the longer the strategy runs. Arbitrage pricing theory excludes doubling strategies for the same reason, since with unlimited credit they would be a money machine (Harrison and Kreps 1979). A martingale-like strategy can be profitable only if the underlying signal has a real edge, and then the sizing rule adds tail risk on top of that edge instead of creating it.
The profile is hidden short volatility: steady income in calm, range-bound markets and large losses in strong trends and crashes. Backtests flatter it, because a short history may contain no move large enough to exhaust the ladder. Goetzmann, Ingersoll, Spiegel and Welch (2007) show that doubling-type strategies can inflate the Sharpe ratio and other performance measures without skill. Brown and Steenbeek (2001) show that Nick Leeson’s trading, which brought down Barings Bank in 1995, followed a doubling strategy. Signs to check are a win rate close to 100%, a Maximum Adverse Excursion (MAE) that grows with position size, and the distance to liquidation at the deepest ladder level. See Kelly criterion for sizing that shrinks after losses and anti-martingale strategy for the opposite approach.
Vaults using this approach: martingale-like, grid trading and mean reversion vaults.
Literature and references:
Doob, Stochastic Processes, Wiley, 1953.
Lester E. Dubins and Leonard J. Savage, How to Gamble If You Must: Inequalities for Stochastic Processes, McGraw-Hill, 1965.
Michael Harrison and David M. Kreps, Martingales and Arbitrage in Multiperiod Securities Markets, Journal of Economic Theory 20(3), 1979.
Stephen J. Brown and Onno W. Steenbeek, Doubling: Nick Leeson’s Trading Strategy, Pacific-Basin Finance Journal 9(2), 2001.
William Goetzmann, Jonathan Ingersoll, Matthew Spiegel and Ivo Welch, Portfolio Performance Manipulation and Manipulation-proof Performance Measures, Review of Financial Studies 20(5), 2007.
See also