Data & Analytics
Backtesting
Also known as:
Historical Testing
Walk-Forward Validation
Updated: 2/11/2026
Validation of a forecasting model on historical data to estimate out-of-sample performance.
Quick Summary
Backtesting validates forecast models with time-based cross-validation – essential against overfitting and look-ahead bias.
Explanation
Time series cross-validation: Train up to t, test on t+1...t+h, slide window. Prevents look-ahead bias.
Marketing Relevance
Without proper backtesting, forecasting results are potentially misleading.
Common Pitfalls
Look-ahead bias. Survivorship bias. Overfitting to backtesting results.
Origin & History
From finance (1990s). Time Series CV popularized by Hyndman & Athanasopoulos.
Comparisons & Differences
Backtesting vs. Cross-Validation
Standard CV shuffles randomly; backtesting respects temporal order.