Where the danger is concentrated

Our risk review covers all 1,515 EAs in the EA Arena. 88 of them are rated "dangerous". Most of those come from two small categories:

  • Martingale: 34 of 40 EAs (85%) rated dangerous.
  • Grid: 14 of 33 (42%).
  • Trend following, momentum, breakout and mean reversion: 2–4% each.

Martingale and grid EAs are 73 of 1,515 EAs (5%) but 48 of the 88 dangerous ones (55%).

How a martingale works — and fails

A martingale EA increases the position after every loss, usually doubling it, so that one win recovers everything lost before. Most of the time this works, which is why these EAs show high win rates and smooth equity curves. The problem is the losing streak. If the first trade risks 0.5% of the account and every loss doubles the next position:

  • after 4 losses in a row the account is down 7.5%;
  • after 6 losses, 31.5%;
  • after 7 losses, 63.5%;
  • the eighth trade alone would then risk 64% of the starting balance, with only 36.5% left. One more loss and the account is gone.

How common are long losing streaks?

We looked at the longest losing streak in 1,095 reliable EA backtests that made at least 20 trades over one year of hourly data (May 2025 to May 2026). The median longest streak was 6 losses. More than a fifth of them (22.5%) had 8 or more losses in a row, and 5.2% had 10 or more. These are ordinary EAs, not martingales — they simply show how often a run of losses happens. For a doubling martingale, a streak that is unremarkable for a normal EA is fatal.

Grids: the same problem in a trend

A grid EA places orders at fixed price steps and adds positions as price moves against it, expecting it to come back. In a range that works. In a trend the positions pile up: over our test year gold rose from about $3,280 to $4,730, so a grid adding one short position every $20 would have opened more than 70 of them, all losing at once. Grids without a maximum number of positions or a stop for the whole basket are the ones our review marks as dangerous.

Why our backtests understate the risk

Our automatic conversion trades a fixed lot size, so it does not reproduce lot doubling or unlimited grid levels. That is why martingale EAs in our backtests show ordinary results (median win rate 49%) instead of the typical "95% wins, then a wipe-out". Treat their backtest numbers as a lower bound on risk and read the risk review on each EA page.

Before you run one

  • Check the maximum lot multiplier and the maximum number of grid levels. If there is no cap, the worst case is the whole account.
  • Work out how many losses in a row your account can absorb at the EA's settings, and compare it with the losing streaks above.
  • Look for a stop for the whole basket of positions, not just per trade.
  • Test in the MT5 Strategy Tester over several years, including fast trends — one quiet year will make almost any martingale look good.