Exponential Smoothing Forecast Calculator
Exponential smoothing is the workhorse of statistical demand forecasting and sits behind most ERP and planning system forecast models. This calculator shows the one step update for simple smoothing and for Holt linear trend smoothing, with alpha and beta you can change live.
Your numbers
Used by the Holt forecast only. Enter 0 for a flat series.
Higher alpha reacts faster to the latest actual.
Your results
This updates one step. To fit alpha and beta properly, test candidate values against your own history and choose those that minimise error. Seasonal items need a seasonal method or deseasonalised data.
Forecast vs alpha
Sweeps Alpha (level smoothing) from 0.05 to 0.95, holding your other inputs fixed.
Compare statistical models on your data
Optional: we can outline a backtest of smoothing and other models on your ERP demand history.
Optional. The tool and every export work without this. No spam.
How to use the calculator
Enter the previous level, which for simple smoothing is just last period's forecast, plus the previous trend and the latest actual. Move the alpha and beta sliders to see the effect. The simple forecast is alpha times the latest actual plus one minus alpha times the previous level. The Holt forecast updates the level, then updates the trend from the change in level, and adds them. The chart sweeps alpha from 0.05 to 0.95 so you can see how strongly the forecast follows the last actual.
The formulas
Simple smoothing: F(t+1) = alpha * A(t) + (1 - alpha) * F(t). Holt: L(t) = alpha * A(t) + (1 - alpha) * (L(t-1) + T(t-1)); T(t) = beta * (L(t) - L(t-1)) + (1 - beta) * T(t-1); forecast = L(t) + T(t). The weights on older data decay geometrically, which is why it is called exponential. A high alpha, such as 0.5 to 0.9, gives a responsive but noisy forecast. A low alpha, such as 0.05 to 0.2, gives a stable forecast that is slow to react.
- Stable, mature items: alpha around 0.1 to 0.3.
- New or volatile items: alpha around 0.3 to 0.6.
- Trended items: use Holt, with beta usually lower than alpha.
Worked example
With a previous level of 500, trend of 8, actual of 520, alpha 0.3 and beta 0.1, the simple forecast is 0.3 x 520 + 0.7 x 500, which is 506. The Holt level is 0.3 x 520 + 0.7 x 508, which is 511.6. The trend becomes 0.1 x 11.6 + 0.9 x 8, about 8.36, and the Holt forecast is about 520. The simple method lags the rising series, which is exactly what Holt corrects.
Common mistakes
Choosing alpha by gut feel and never revisiting it is the most common problem; refit regularly using a holdout. Using a high alpha to chase every spike produces a nervous forecast that causes planning churn. Applying simple smoothing to a trended or seasonal series produces consistently biased forecasts, so pair it with the tracking signal. Finally, start values matter for short histories, so initialise the level from the average of the first several periods.
Frequently Asked Questions
What is a good alpha for exponential smoothing?
There is no universal value. Stable demand usually works with 0.1 to 0.3, volatile or changing demand with 0.3 to 0.6. The best way is to try several alphas on past data and choose the one with the lowest error, then recheck periodically.
What is the difference between simple and Holt smoothing?
Simple smoothing tracks a level only, so it lags when demand trends up or down. Holt adds a second equation for trend, so it can project continuing growth or decline. Use Holt when you see a sustained direction, and simple smoothing for flat series.
How is exponential smoothing different from a moving average?
A moving average weights the last n periods equally and ignores older data. Exponential smoothing weights every past period with geometrically decreasing weights, so it needs only the last forecast and the latest actual to update. Both are baselines for stable demand.
Can exponential smoothing handle seasonality?
Not the simple or Holt versions. Holt-Winters adds seasonal factors. A practical alternative is to deseasonalise demand with a seasonal index, forecast the deseasonalised series, then reapply the index to the result.
If your ERP or planning tool already runs smoothing models, we can review the parameters it uses and help compare them with alternatives on your own history.
Related Tools
Moving Average Forecast Calculator
Enter the last six periods of demand, choose a 3, 4 or 6 period window and get the simple moving average forecast with a comparison to the naive forecast.
Supply Chain & InventorySeasonal Index Calculator and Worksheet
Calculate the seasonal index for a month from three years of history and apply it to next year's annual forecast to get the monthly forecast.
Supply Chain & InventoryForecast Bias and Tracking Signal Worksheet
Enter actual and forecast for each period of one item and get running error, MAD, tracking signal and bias percent so you know when a forecast has drifted.
Infor M3 & CloudSuiteM3 Exponential Smoothing Forecast Calculator (Holt Trend)
Step a Holt level and trend forecast forward one period and see how the smoothing constant changes the result.
ERP OperationsAI Demand Forecasting Improvement Calculator
Turn a forecast error reduction into freed-up working capital and reduced stockout losses, based on your revenue, inventory value, and carrying cost.
More Supply Chain & Inventory tools
See all- Oracle Fusion Inventory Turns and Carrying Cost Calculator
- Distribution ERP Fill Rate Calculator
- Distributor Rebate Tracking Worksheet
- Branch Transfer Planning Worksheet
- Distributor Slow-Moving Inventory Review Worksheet
- Distributor Vendor Performance Scorecard Worksheet
- Distributor GMROI Calculator
- Distributor Backorder Cost and Fill Rate Calculator
- EOQ with Vendor Minimum Order Calculator