Exponential Smoothing Forecast Calculator

Period	Actual	Smoothed
1	100	100
2	105	101.5
3	102	101.65
4	108	103.55
5	110	105.49
6	107	105.94
7	112	107.76

Period

Actual

Smoothed

100

105

101.5

102

101.65

108

103.55

110

105.49

107

105.94

112

107.76

About the Exponential Smoothing

The Exponential Smoothing Forecast Calculator produces a next-period forecast from a historical time series using simple exponential smoothing, a method that weights recent observations more heavily than older ones. It is a lightweight alternative to a moving average when you want every past data point to still contribute something to the forecast.

How It Works

You enter a series of historical values in order and a smoothing constant α between 0.01 and 1. The calculator sets the first smoothed value equal to the first observation, then works forward, blending each new actual value with the previous smoothed value according to α, and reports the final smoothed value as the forecast for the next period along with a period-by-period table.

Smoothed[1] = Actual[1]; Smoothed[t] = alpha * Actual[t] + (1 - alpha) * Smoothed[t-1] for t > 1; Forecast = Smoothed[last period]

Formula & Methodology

Start by setting the smoothed value for period 1 equal to the first actual observation. For every following period, multiply the current period's actual value by α and multiply the previous period's smoothed value by (1 - α), then add the two together to get the new smoothed value. Repeat through the last data point; that final smoothed value is the one-period-ahead forecast, since simple exponential smoothing carries the last smoothed level forward unchanged.

Examples

Weekly demand series

For the values 100, 105, 102, 108, 110, 107, 112 with alpha set to 0.3, the smoothed series ends at approximately 105.85, so the forecast for the next period is about 105.85 even though the most recent actual value was 112.

Higher responsiveness setting

Using the same 7-period series but raising alpha to 0.7 pulls the smoothed series much closer to recent actuals, producing a next-period forecast noticeably closer to the last few observed values than the alpha 0.3 case.

Advantages

Produces a forecast from the full historical series, letting older data still contribute a diminishing influence rather than being dropped entirely
Lets you directly control responsiveness versus smoothness by adjusting a single parameter, α
Returns a full period-by-period table of smoothed values alongside the forecast, making it easy to see how the smoothing evolved over time

Common Mistakes

Choosing an alpha value arbitrarily rather than testing a few values and comparing which tracks the historical series most accurately
Applying simple exponential smoothing to data with a strong trend or seasonal pattern, which this method does not model and will systematically lag behind
Treating the single forecast value as reliable far beyond one period ahead, when simple exponential smoothing only produces a flat forecast for all future periods

Edge Cases to Watch For

At least 2 historical values are required, or the calculator returns an error rather than a one-point forecast.
The smoothing constant must be greater than 0 and no greater than 1; values of 0 or below, or above 1, are rejected.
A high α close to 1 makes the forecast track the most recent value very closely and react quickly to changes, while a low α close to 0 produces a much smoother series that lags behind sudden shifts.

Common Use Cases

Inventory and demand planners generating a short-term forecast from recent sales or usage history
Analysts comparing how different smoothing constants affect forecast responsiveness before choosing one for ongoing use
Students learning the mechanics of exponential smoothing as a stepping stone to more advanced forecasting methods like Holt-Winters

Frequently asked questions

How does simple exponential smoothing choose its forecast?

It's a weighted average of all past observations, with weights that decay exponentially the further back in time you go - unlike a simple moving average, which weights all included points equally and ignores older ones entirely, exponential smoothing lets every past observation contribute, just with diminishing influence.

Conclusion

Simple exponential smoothing gives a quick, adjustable forecast that weights recent history more heavily without discarding older observations. Because it produces a flat forecast and ignores trend or seasonality, it works best for relatively stable series or as a baseline before trying more advanced methods.