Overfitting and walk-forward testing

When a strategy only works on the past

Professional - Gold & Platinum

In this lesson

Recognise overfitting and use out-of-sample and walk-forward testing to check that a strategy is robust.

Key takeaways

• Overfitting = fitting noise in past data; it fails on new data.
• Keep an out-of-sample period you use only once, for judging.
• Walk-forward testing repeatedly tunes on the past and tests on the next period.
• Robust strategies work across neighbouring settings (a plateau), not one lucky value.

Next lesson: Monte Carlo: what could go wrong

Educational information only - not financial advice.