Holt–Winters Exponential Smoothing: Forecasting When Patterns Repeat

Explaining a Simple Time Series Model

The first time you try exponential smoothing on a seasonal signal, it can feel disappointing.

You smooth the data.
You forecast into the future.
And the forecast comes out flat.

Nothing is wrong. The model is doing exactly what you asked. But what you asked for was too simple for the problem.

This article explains why that happens, and how the Holt–Winters a…