Calculator inputs
Example data table
This example shows a short monthly series suitable for checking forecast bias and control-limit behavior.
| Period | Actual | Forecast | Actual − Forecast | Absolute Error |
|---|---|---|---|---|
| Jan | 120 | 118 | 2 | 2 |
| Feb | 128 | 130 | -2 | 2 |
| Mar | 125 | 127 | -2 | 2 |
| Apr | 134 | 129 | 5 | 5 |
| May | 138 | 135 | 3 | 3 |
| Jun | 142 | 144 | -2 | 2 |
Formula used
Forecast Error (FE) = Actual − Forecast, or Forecast − Actual if your team uses that sign convention.
RSFE = ΣFE, the running sum of forecast errors through the current period.
MAD = Σ|FE| ÷ n, where n is the number of analyzed periods.
Tracking Signal = RSFE ÷ MAD.
Control check: if |Tracking Signal| exceeds the selected control limit, the forecast process is considered out of control. If it exceeds the warning limit only, review the series before a full breach occurs.
How to use this calculator
- Enter a scenario name so exports and summaries stay organized.
- Paste the actual series and matching forecast series in the same order.
- Optionally add period labels such as months, weeks, or SKU review cycles.
- Choose the forecast error sign convention used by your team.
- Set a control limit and an earlier warning threshold.
- Press the calculate button to show the result above the form.
- Review RSFE, MAD, tracking signal, warnings, breaches, and bias direction.
- Use the CSV or PDF buttons to save the analysis for meetings or audits.
Frequently asked questions
1. What does tracking signal measure?
Tracking signal measures cumulative forecast bias relative to average absolute error. It helps determine whether repeated forecast errors show a consistent directional problem instead of random variation.
2. Why are ±4 limits common?
Many planning teams use ±4 as a practical rule for identifying meaningful bias without overreacting to ordinary noise. The best limit still depends on process stability, service goals, and review policy.
3. What is RSFE?
RSFE means Running Sum of Forecast Errors. It accumulates signed forecast errors over time and reveals whether forecasts are systematically high or low across consecutive periods.
4. Why does MAD matter in tracking signal?
MAD scales cumulative bias by typical error size. Without MAD, RSFE alone can look large simply because the numbers are large, not because the forecasting process is badly biased.
5. Can I use weekly or daily data?
Yes. The calculator works with any evenly sequenced time periods, including daily, weekly, monthly, or quarterly series, as long as actual and forecast values align correctly.
6. What does a positive tracking signal mean?
Its meaning depends on your selected error formula. With Actual − Forecast, a positive value usually points to under-forecasting. With Forecast − Actual, a positive value suggests over-forecasting.
7. Should I act on a warning before a breach?
Usually yes. A warning zone can reveal drift before the process crosses the main control limit. Early review can reduce service disruptions, excess stock, or recurring forecast overrides.
8. Is this enough for full forecast evaluation?
No. Tracking signal is useful for bias control, but it should be reviewed alongside MAPE, MAD, RMSE, service levels, and business context for a more complete forecast assessment.