Exponential Smoothing / ETS / Holt-Winters

Best for: Trend and seasonality Aliases: ETS, Holt-Winters

How it works

Decomposes the series into level $l_t$, trend $b_t$, and seasonal $s_t$ components updated by exponentially weighted recursions, e.g. $l_t=\alpha(y_t-s_{t-m})+(1-\alpha)(l_{t-1}+b_{t-1})$ and $b_t=\beta(l_t-l_{t-1})+(1-\beta)b_{t-1}$. The $h$-step forecast is $\hat y_{t+h|t}=l_t+hb_t+s_{t+h-m(k+1)}$, where $m$ is the season length and $k=\lfloor(h-1)/m\rfloor$. Smoothing parameters $\alpha,\beta,\gamma$ control how fast each component adapts; additive or multiplicative variants handle constant vs. growing seasonal amplitude.

When to use

Business forecasting with trend and seasonality where you want a robust, interpretable baseline fast.

Watch out

Mostly univariate; assumes a stable seasonal pattern; can lag through regime shifts.

Common fields

Retail · inventory · operations