Risk Management
Position sizing, stops, and surviving the bad path.
Size from drawdown, not return
The most common mistake is sizing a position from expected return. Return is the least reliable number in any backtest; drawdown is far more stable across regimes.
Take the P95 drawdown from Monte Carlo, decide the maximum account loss you can tolerate without abandoning the strategy, and let those two numbers set your position size.
Stops and their cost
A tight stop reduces per-trade loss but increases the number of trades stopped out on noise, which raises total fees paid and can invert an edge. The engine reports fees paid as a first-class metric precisely so you can see this trade-off.
Test stop levels as a parameter sweep like any other. If your strategy only works at exactly a 1.8% stop, it does not work.
Strategies with more than roughly 30% of exits triggered by stops usually have an entry-timing problem, not a stop problem.
Correlation across strategies
Running five strategies that all buy momentum on major pairs is running one strategy at five times leverage. Check that your deployed set does not share the same underlying condition.
The Paper Bots view shows aggregate portfolio equity alongside each individual bot so correlated drawdowns become visible early.