Calculate demand forecasts using simple exponential smoothing. Apply a smoothing constant (alpha) to weight recent demand more heavily.
Simple Exponential Smoothing (SES) forecasts next-period demand using a weighted combination of the most recent actual demand and the previous forecast. The smoothing constant alpha (α) controls how much weight is given to the latest observation versus the historical forecast.
When α is close to 1, the forecast reacts quickly to recent changes. When α is close to 0, the forecast changes slowly and smooths out noise. SES inherently applies exponentially decreasing weights to older observations, giving it more responsiveness than a simple moving average.
This calculator lets you enter a sequence of demand values and an alpha value to compute the exponentially smoothed forecast for the next period step by step.
Supply-chain managers, warehouse operators, and shipping coordinators rely on precise exponential smoothing forecast data to maintain efficiency and control costs across complex distribution networks. Revisit this calculator whenever conditions change to keep your logistics plans aligned with real-world performance.
Exponential smoothing is the workhorse of demand planning — simple to implement yet effective for a wide range of demand patterns. It requires only one parameter (α) and minimal historical data. This calculator makes it easy to experiment with different alpha values to find the best fit for your demand profile.
F_t = α × D_{t-1} + (1 − α) × F_{t-1} Where: F_t = Forecast for period t D_{t-1} = Actual demand in the previous period F_{t-1} = Previous forecast α = Smoothing constant (0 < α < 1)
Result: Next Period Forecast = 109.7
Starting with F1 = 100: F2 = 0.3×100 + 0.7×100 = 100. F3 = 0.3×110 + 0.7×100 = 103. F4 = 0.3×105 + 0.7×103 = 103.6. F5 = 0.3×120 + 0.7×103.6 = 108.5. F6 = 0.3×115 + 0.7×108.5 = 110.5.
In SES, the most recent observation gets weight α, the one before gets α(1−α), the one before that gets α(1−α)², and so on. Weights decrease exponentially, meaning very old data has negligible influence. This self-weighting property makes SES adaptive without explicitly choosing a window size.
Optimal alpha can be found by minimizing the sum of squared errors over historical data. Many tools use grid search (testing alpha from 0.01 to 0.99 in steps of 0.01) to find the minimum-error value. In practice, alpha between 0.1 and 0.3 works well for most supply chain applications.
SES is the foundation of a family of methods: Holt's method adds a trend component (double exponential smoothing), and Holt-Winters adds seasonality (triple exponential smoothing). Understanding SES provides the basis for these more sophisticated approaches.
Alpha (α) is a value between 0 and 1 that determines how much weight the model gives to the most recent demand observation. Higher alpha values make the forecast more responsive; lower values produce smoother, more stable forecasts.
Start with 0.2 and adjust. Fit the model to historical data using different alpha values and select the one that minimizes forecast error (MAD or MSE). Most demand planning software can optimize alpha automatically.
SMA gives equal weight to all N past periods. SES applies exponentially decreasing weights, emphasizing recent data more. SES typically responds faster to demand changes and requires only one parameter instead of choosing N.
No. Simple exponential smoothing assumes level demand. For seasonal patterns, use Holt-Winters (triple exponential smoothing), which adds seasonal components to the model.
Common approaches include using the first actual demand value, the average of the first 3–5 periods, or a known baseline forecast. The initialization has less impact as more periods are processed.
Yes. SES remains widely used because of its simplicity, interpretability, and low data requirements. It often performs comparably to complex models for short-term forecasting of stable demand patterns.