Skip to content
ML Academy

Introduced on day 24

Recursive forecasting

Multi-step forecasting with one one-step model whose predictions are fed back in as inputs for the next step, so errors compound with the horizon.

In detail

Only the first forecast uses observed values for its newest lags; each later step reads earlier predictions. The strategy needs future-known or forecast values for every other input at each step, and its backtest must feed predictions back exactly as production does. Classical ETS and ARIMA forecasts are produced this way natively.