Exponential Smoothing Forecast Calculator
Simple exponential smoothing of a time series — next-period forecast for your α, the smoothed series, MAE/MAPE and the best α found by grid search.
How it works
Fₜ₊₁ = α·Yₜ + (1 − α)·Fₜ, F₁ = Y₁; weights decay as α(1−α)ᵏ over past observations
Simple exponential smoothing forecasts the next value as a weighted average of all past observations with geometrically fading weights. A high α reacts quickly and follows noise; a low α is smooth but slow. It suits level series without strong trend or seasonality — for trending data like the example, forecasts lag and a higher α or Holt’s double smoothing does better.
Worked example
Twelve months of sales ending at 195 with α = 0.3: forecast 178 for next month, MAE 13.7. Grid search prefers α = 1.0 (a naive last-value forecast) because the series trends steadily.
Frequently asked questions
How do I pick α?
Minimise a forecast error such as MAE or MSE on past data — the grid search above — or start at 0.2–0.3 for stable series.
What about trend and seasonality?
Use Holt (trend) or Holt–Winters (trend + seasonality); simple smoothing will systematically lag a trend.
Related calculators
This calculator is for educational purposes. Double-check important results before acting on them.