SONIQE
Risk metric

What Is a Good Drawdown for a Trading Bot?

There is no universal percentage that makes a trading bot safe. Drawdown becomes meaningful when compared with return, leverage, strategy behaviour and how much evidence exists.

Lower is generally easier to survive

All else equal, a strategy that achieves a given return with lower drawdown has exposed capital to less observed peak-to-trough loss. But “all else equal” rarely holds across different systems.

Return and drawdown belong together

A 5% drawdown accompanying a modest annual return means something different from the same drawdown alongside very high monthly returns. Ask what risk produced the return rather than judging either number alone.

Short histories can look deceptively smooth

A strategy may simply not have encountered its difficult market regime yet. Historical maximum drawdown usually becomes more informative as the live record grows.

Leverage changes the consequence

Amplification or high position sizing can make a small underlying strategy drawdown economically large relative to contributed capital.

Stress beyond the historical maximum

When sizing an allocation, model losses larger than the worst previously observed drawdown. History is evidence, not a protective boundary.

Trading drawdown explained →

Calculate recovery after a loss →