A trader sends you a backtest report. The equity curve is almost perfectly smooth — barely any drawdown, 95% win rate, $50,000 profit on a $10,000 account. It looks too good to be true.
It probably is. The EA is almost certainly a martingale.
Martingale strategies are the single most common cause of retail account blowups in forex. They produce gorgeous backtests because they’re mathematically designed to win almost every trade. The “almost” is where the destruction lives — and most traders never see it coming because the backtest doesn’t show it.
What Martingale Actually Means in EA Trading
In the strict mathematical sense, a martingale is a betting strategy that doubles the stake after every loss, so that the first win recovers all previous losses plus a profit equal to the original stake. In EA trading, the definition is broader. Any strategy that systematically increases risk after a loss — whether by doubling lot size, adding positions (grid), or holding losers while averaging down — exhibits martingale behavior.
The common variations you’ll encounter:
Classic lot doubling: After a loss, the next trade uses 2x the lot size. After two consecutive losses, 4x. After three, 8x. The math is simple — eventually a win occurs and recovers everything. Until the losing streak exceeds available margin, at which point the account can be wiped out in a single session.
Grid strategies: The EA opens multiple positions at predetermined intervals as price moves against the initial trade. Each new position “averages” the entry price, so a smaller reversal is needed to close profitably. The problem: during a strong trend, the EA accumulates massive exposure in a losing direction.
Recovery mode EAs: After a losing trade, the EA enters a “recovery” sequence with larger position sizes or tighter take-profits designed to recoup the loss quickly. This is martingale with extra steps.
Hidden averaging: The EA doesn’t explicitly double lot sizes, but it holds losing trades significantly longer than winning trades — effectively letting losers accumulate unrealized loss before eventually closing them in profit. The backtest shows a high win rate, but each “win” was at one point a deep unrealized loser.
Why Backtests Hide the Risk
A martingale EA backtest typically shows 80-99% win rate with a smooth, upward-sloping equity curve. The maximum drawdown appears contained — maybe 15-20%. This looks attractive compared to trend-following strategies with 40-50% win rates and choppy equity curves.
The deception is a survivorship artifact. The backtest covers a fixed historical period. If no losing streak exceeded the account’s margin capacity during that period, the martingale “worked.” But the backtest can’t show what would have happened with a slightly longer losing streak, a wider spread day, or a flash crash during an accumulation sequence.
The statistical distribution of martingale outcomes is heavily skewed: many small wins and rare, catastrophic losses. Extending the backtest by a few months — or testing on a slightly different period — often reveals the hidden blowup event. But sellers of martingale EAs carefully choose backtest periods that avoid these events.
This is exactly why automated detection matters. You can upload any EA backtest report to our free analyzer and get an instant martingale safety score.
The MG Indicator: A Statistical Detection Method
You can detect martingale behavior without reading the EA’s source code. The method relies on a simple statistical ratio that examines the relationship between winning trades and what happened before them.
Calculate two values:
Average win after a loss (WAL): Look at every winning trade that was immediately preceded by a losing trade. Calculate the average profit of these wins.
Average win after a win (WAW): Look at every winning trade preceded by another winning trade. Calculate the average profit.
The MG Indicator is the ratio: WAL ÷ WAW
For a clean, non-martingale EA, this ratio should be close to 1.0. Wins are approximately the same size regardless of what happened on the previous trade. The lot size is consistent, and there’s no recovery mechanism inflating post-loss trades.
Here’s what the ratio tells you:
Below 1.2 — Clean: Win sizes are consistent. No evidence of position sizing manipulation after losses. The EA treats each trade independently.
1.2 to 1.5 — Minor bias: Wins after losses are slightly larger. This could be genuine (e.g., a volatility-based sizing system that naturally increases after drawdown periods) or it could be soft martingale. Worth investigating but not necessarily dangerous.
1.5 to 2.5 — Recovery trading: Wins after losses are 1.5x to 2.5x larger than normal wins. The EA is clearly doing something different after a loss. This is the danger zone — the strategy may survive for months or years before an extended losing streak overwhelms it.
Above 2.5 — Martingale confirmed: Post-loss wins are more than 2.5x normal wins. The EA is aggressively increasing position size or holding losers until they become winners. The risk of account destruction is extremely high.
Other Warning Signs
Beyond the MG Indicator, watch for these patterns in backtest reports:
Win rate above 85%: Very few legitimate trading strategies achieve this consistently. Most strategies with genuine edge operate between 40-70% win rate. An extremely high win rate almost always means the EA is taking small profits while letting losers run — martingale’s signature.
Average loss much larger than average win: A risk-reward ratio heavily skewed toward losses (e.g., average win $20, average loss $180) suggests the strategy is harvesting small wins while occasionally taking catastrophic losses. This is the mathematical fingerprint of martingale — asymmetric payoff with negative expected tail risk.
Maximum consecutive losses is very low (1-2): In any trading strategy with a reasonable number of trades, you should see consecutive losing streaks proportional to the loss rate. A 60% win rate over 200 trades should produce at least one streak of 4-5 consecutive losses. If the maximum consecutive losses is suspiciously low, the EA may be holding losers open until they become winners — turning “consecutive losses” into “one big loss eventually recovered.”
Holding time asymmetry: If losing trades are held 2-3x longer than winning trades on average, the EA is likely waiting for losers to recover rather than cutting losses. This is one of the most reliable martingale indicators and one of the 17 tests in Edge Matrix’s validation suite.
What to Do If You Detect Martingale
If the MG Indicator is above 1.5, your options are limited:
Walk away. This is the correct answer in almost every case. The expected long-term outcome of a martingale strategy, given enough time, trends toward account destruction. No amount of “risk management” changes the fundamental math — you’re just extending the timeline.
If you still want to trade it: Size it at no more than 5% of your total portfolio. Accept that this allocation could realistically go to zero. Calculate the maximum theoretical drawdown (which for a pure martingale approaches 100%) and ensure you can absorb it without affecting your overall portfolio.
If you’re the developer: Examine why post-loss trades are larger. If it’s an intentional recovery mechanism, consider removing it and accepting the lower win rate. A strategy that profits without martingale behavior is vastly more valuable than one that requires it — because it can actually survive long enough to compound.
Test Any EA Right Now
Our free Monte Carlo analyzer includes automatic martingale detection. Upload any MT4 or MT5 backtest report and the MG Indicator is calculated as part of the robustness analysis. You’ll see the raw ratio and a 0-40 safety score — 40 for clean trading, 0 for confirmed martingale. It runs entirely in your browser, so your strategy data stays private.
Don’t trust equity curves. Trust statistics.