← All methods

ETS · Pro

ETS auto

Pick the best of SES / Holt / damped / Holt–Winters on SSE, then backtest (Pro).

How it works

ETS auto is a Hyndman-style family search over the smoothing methods already in the Free catalog. It picks on in-sample SSE, then the usual rolling-origin MASE ranks that pick against everyone else — including Croston on intermittent data, where ETS is often the wrong family. It is Pro convenience, not a new process.

Sample forecast question

Same 12-quarter café series as additive Holt–Winters. Which ETS candidate wins the in-sample grid, and what is the next-year forecast path?

Step-by-step on these numbers

1. The question

   Same 12-quarter café series as additive Holt–Winters. Which ETS candidate wins the in-sample grid, and what is the next-year forecast path? Sample quarters: Y1 Q1=40, Y1 Q2=48, Y1 Q3=44, Y1 Q4=55, Y2 Q1=42, Y2 Q2=51, Y2 Q3=46, Y2 Q4=58, Y3 Q1=45, Y3 Q2=53, Y3 Q3=48, Y3 Q4=61 (covers).

2. Inner grid

   ETS auto fits SES, Holt, and damped trend; with m > 1 it also tries Holt–Winters (multiplicative only if y > 0). It picks the lowest in-sample SSE, then the add-in still ranks that pick with rolling-origin MASE against Croston and everyone else.

3. Winner on this sample

   The inner grid chose Holt–Winters (seasonal). Parameters: alpha = 0.400, beta = 0.050, gamma = 0.500, level = 52.753, trend = 0.578.

4. Forward path

   The engine’s ETS forecast uses ŷ_{n+h} = ℓ + h·b with the winner’s level/trend (seasonal indices, if any, are in the DNA of the source method). Y4 Q1: 53.33; Y4 Q2: 53.91; Y4 Q3: 54.49; Y4 Q4: 55.06.

5. Interval

   Residual σ = 0.895. For h = 1 the 95% band is [51.58, 55.09] around 53.33. The engine labels this interval native for the method.

Graph of this sample

Related methods: Simple exponential smoothingHolt linearHolt–Winters additive

Educational only. Not investment, weather, or operational advice. In the add-in, rank is rolling-origin MASE — “best supported among candidates on this series.”

Use it in the Excel Add-in