Grid/DCA: why accounts blow up — and the 4 mandatory safety limits

Grid/DCA is the most addictive strategy family for new bot traders: 90%+ win rate, silky equity curve, green every month. Until one day — usually a big-news day — the account evaporates. Here is why, and if you still want to run it, the four mandatory safety locks.

The mechanic: trading many small wins for one total loss

DCA (averaging down) adds same-direction positions as price moves against you, pulling breakeven closer. Grid places evenly spaced buy/sell orders. Both rest on one belief: price will come back.

In ranging markets — most of the time — that belief holds and the bot wins steadily. The problem is the profit distribution is the inverse of trend-following: many small wins, one enormous loss. A 5% no-retrace trend (gold does this several times a year) wipes six months of profit plus most of the principal, because position size ballooned at exactly the wrong moment.

The "95% win rate" illusion

A grid/DCA's high win rate is not an edge — it is how the strategy postpones losses. Losing trades are never cut; they are "nursed" with new entries. A 1–2 year backtest may simply not contain a trend strong enough to expose the kill zone, producing unrealistically pretty results. That is why grid backtests must cover strong-trend periods (for gold: the 2024–2025 rallies, Fed/NFP news drops).

If you still run it: 4 mandatory locks

1. Max DCA levels (MaxDCAOrders): 3–5. No cap = blow-up is a matter of time.
2. Max total volume (MaxTotalLot): the sum across levels must stay within what you can survive if the global stop is hit.
3. Account-level stop (MaxDrawdownPercent): down 10–15% → close everything, take the loss, live on. This is the most important lock and the one people love to disable "because it keeps cutting at the bottom" — until the day it saves the account.
4. News/session filter: no new grids around major releases (FOMC, NFP, CPI); on gold, also avoid thin early-Asia liquidity.

Straight conclusion

Grid/DCA is not a scam — it is a short-volatility strategy with heavy tail risk, suitable only if you understand that, run it on small ring-fenced capital, and keep all four locks. When you build a grid/DCA bot on Forex AI Lab, the system always adds these limit inputs and a warning in the bot profile — do not turn them off.