Lessons from martingale EAs that failed

[FXS London Scalper]

Over the years I've tested several martingale and grid EAs, partly out of curiosity, partly because the backtests looked amazing. All of them failed eventually. Notes for anyone tempted.

What the backtests showed

• Smooth equity curves, 90%+ win rates.
• Rare but deep drawdowns, often "recovered".

What happened live

1. EA one (EURUSD, doubling after each loss): blew 70% of the account during a strong trend day after five months of profits.
2. EA two (grid on EURCHF): fine until a range broke. The grid kept adding and margin ran out.
3. EA three (gold, "smart martingale" with filters): survived longer, then one news spike triggered all levels at once.

Why

• They exchange many small wins for a rare huge loss. The backtest often doesn't include the worst period.
• Position size grows exactly when the market is most against you.
• Spreads and slippage at the worst moment are worse than in any backtest.

The useful lesson: always look at the worst trade and maximum exposure, not win rate. If one trade can risk over 10% of the account, the system is a bet, not a strategy.

Anyone ever run a martingale long term without blowing up? I'm genuinely curious what the risk controls looked like.