Exponential smoothing

Exponential smoothing is a forecasting method that sets the next forecast as a weighted blend of the latest actual value and the previous forecast, using a smoothing constant α.

ByHoang TruongUpdated July 11, 2026

See it move

Loading infographic...

Exponential smoothing blends the latest actual with the previous forecast. If this month's forecast was €50,000 and the actual came in at €56,000, a smoothing constant of α = 0.3 gives next month's forecast as 0.3 × €56,000 + 0.7 × €50,000 = €51,800 — sitting closer to the old forecast because α is small. Only the previous forecast and the latest actual are needed to update, no full history.

Where it fits

TopicDescriptive StatisticsAdvancedSubjectData Analysis & StatisticsAdvanced

The formula

LaTeX

F_{t+1} = \alpha A_t + (1-\alpha) F_t

Variables

$F_{t + 1}$: New forecast for the next period (€)
$α$: Smoothing constant, between 0 and 1
$A_{t}$: Latest actual observation (€)
$F_{t}$: Previous forecast (€)

Produces the next-period forecast as a weighted blend of the latest actual value and the previous forecast.

Check yourself

PracticeCORE

A retailer's forecast for this month was €80,000, and the actual outcome was €92,000. Using a smoothing constant of α = 0.4, what is the forecast for next month?

Select an answer to check your understanding.