backtest-diagnose

To diagnose a failing backtest, inspect artifacts like metrics.csv, equity.csv, and trades.csv alongside signal_engine.py and config.json to classify runtime errors, logic bugs, or data problems as the root cause.

Backtest abnormal results and NaN values often stem from logic bugs or data problems. Inspecting equity.csv and trades.csv artifacts helps pinpoint the fault domain within your signal engine configuration and code paths.

The best way to fix underperforming backtest results is applying a structured workflow: inspect artifacts, review code, classify the root cause, apply targeted edits, and re-run the backtest to verify metrics converge.

Verify a backtest fix by re-running the test to confirm exit_code equals 0, ensuring no NaN values appear in key series, and checking that performance metrics converge to expected values.

Yes, backtest diagnostics work with custom code by reviewing code paths in files like signal_engine.py and config.json to understand the fault domain and identify logic bugs or runtime errors.

Debugging backtest failures from data artifacts requires complete artifacts like metrics.csv and trades.csv. Limitations arise if artifacts are missing, incomplete, or lack sufficient detail to classify runtime errors versus data problems.