Backtesting Validation

Focus

Measure realistic performance and reduce false positives.

When Not to Use

Do not use for exchange integration bugs or runtime infrastructure incidents unless they affect simulation assumptions directly.

Minimum Standards

• Include all relevant costs: maker/taker fees, borrow/funding where applicable.
• Model slippage and execution delay explicitly.
• Separate in-sample development from out-of-sample evaluation.
• Use walk-forward or rolling windows for non-stationary markets.

Workflow

1. Define simulation assumptions:
   • order types, fill model, latency, fees, and slippage regime.
2. Build split protocol:
   • train/validate/test or anchored/rolling walk-forward.
3. Evaluate metrics:
   • CAGR, Sharpe, max drawdown, Calmar, turnover, hit ratio.