Robustness Metrics
This page explains common metrics used to review backtest and optimization results.
These metrics do not prove that a setup will work live. They provide extra context beyond net profit.
Why Metrics Matter
Two test results can have similar net profit but different risk profiles.
Reviewing drawdown, volatility, and recovery can help you decide which setup deserves further testing.
Sharpe Ratio
Sharpe ratio compares return with volatility.
Use it as a rough risk-adjusted metric. It can be misleading when returns are not normally distributed, when trade count is small, or when a few trades dominate the result.
Calmar Ratio
Calmar ratio compares return with maximum drawdown.
It can help show whether a result required a large drawdown. It is still based on the selected historical period.
Recovery Factor
Recovery factor compares net result with maximum drawdown.
It can help show how much historical gain was produced relative to the largest drawdown in the test.
Composite Fitness
Some BananaEA versions may expose a composite fitness value for optimization review.
Treat it as a sorting aid, not a final decision rule.
After using any composite score:
- Review the underlying metrics.
- Check trade count.
- Check drawdown.
- Test out-of-sample.
- Forward test on demo.
How To Use These Metrics
Use metrics to ask better questions:
- Was the result driven by one trade?
- Was drawdown acceptable?
- Did nearby settings behave similarly?
- Did the setup survive out-of-sample testing?
- Were broker assumptions realistic?
What Not To Do
Do not:
- Choose a setup from one metric alone.
- Treat a high score as live readiness.
- Ignore spread, commission, and slippage.
- Ignore low trade count.
- Re-optimize until the score looks attractive.