Back to library
Library term·Algorithmic trading

Avoiding Curve Fitting & Over-Optimisation

Too many free parameters + short history = beautiful backtest, fragile live.

Authored by·Editorially reviewed
Onur Erkan Yıldız
Founder, Financial Engineer · CMB-licensed

Overview

Defences: parsimony, cross-validation, synthetic perturbations, Monte Carlo path reshuffles.

Practical takeaway

If Sharpe in live << Sharpe in test with similar volatility, suspect implementation bug or over-fit.

How this connects to Finvestopia

Finvestopia philosophy: disclose uncertainty — same ethos as our confidence bands on radar statistics.

Related entries

Monte Carlo Stress Tests for Trading Systems

Shuffle trade order, perturb returns, bootstrap paths to estimate drawdown distributions.

Walk-Forward Optimisation

Train parameters on in-sample windows, validate on subsequent out-of-sample slices — fights curve fitting.

Educational content authored by our team — informational only, not investment advice.