← AI & ML glossary
Definition

Forecast backtesting

Repeatedly training on historical cutoffs and evaluating later periods to estimate how a forecasting system will behave in production.

Why this matters in practice

The useful question is rarely whether a team can use the term. It is whether the method changes a decision, survives real data, respects security constraints, and can be monitored after launch.

Browse technical explainers →