A forecast is only useful if you know how wrong it tends to be. Compare actuals with forecasts to measure accuracy, detect systematic bias, and flag forecasts that need recalibration using the tracking signal.
How it works
Measure how accurate and how biased a forecast has been: percentage errors, absolute and squared errors, bias and the tracking signal used to detect a drifting forecast.
Error = Forecast − Actual
MAPE = average |Error ÷ Actual|; WMAPE = Σ|Error| ÷ Σ|Actual|
MAD = average |Error|; RMSE = √(average Error²)
Bias = average Error; Tracking signal = Σ Error ÷ MADCommon uses
- Monthly forecast performance reviews
- Comparing forecasting methods and models
- Demand planning and S&OP processes
Worked example
The calculator above opens with this example. Change any input to see the results update instantly, or select Reset to return to these figures.
| Actuals | 1,248 1,151 1,280 1,266 1,319 1,305 1,272 1,298 1,341 1,377 1,352 1,420 |
| Forecasts | 1,210 1,190 1,240 1,250 1,280 1,300 1,290 1,285 1,310 1,330 1,340 1,365 |
| MAPE | 2.3% |
| Weighted MAPE (WMAPE) | 2.3% |
| Mean absolute deviation (MAD) | 29.42 |
| RMSE | 33.16 |
| Bias (mean error) | -19.92 |
| Tracking signal | -8.12 |
The tracking signal is outside ±4: the forecast is systematically biased and should be recalibrated. Error = Forecast − Actual, so a positive bias means the forecast ran high.
Interpreting the results
- WMAPE is more reliable than MAPE when some actuals are small, because small denominators inflate MAPE.
- RMSE penalizes large misses more heavily than MAD; use it when big errors are costly.
- Accuracy shows how far off the forecast is; bias shows whether it is consistently high or low. Track both.
Frequently asked questions
What is a good MAPE?
Benchmarks vary by industry and level of detail. As a rough guide, a MAPE under 10% is very good, 10% to 20% is good, and above 20% suggests room for improvement. Item-level forecasts naturally run higher than totals.
What is the tracking signal?
The tracking signal is cumulative forecast error divided by the mean absolute deviation. A value outside roughly ±4 indicates the forecast is consistently too high or too low and should be investigated.
Why use WMAPE instead of MAPE?
MAPE can be inflated by periods with very small actual values, because a small denominator magnifies the percentage error. Weighted MAPE divides total absolute error by total actuals, so large periods carry more weight.
Related calculators
- Moving Average & Exponential Smoothing Forecast: SMA, WMA, simple, Holt and Holt-Winters forecasts with fit statistics.
- Rolling Forecast Builder: 12- to 24-month rolling forecast from monthly actuals using four methods.
- Budget vs. Actual Variance Analyzer: Line-item variances with favorable/unfavorable flags and investigation thresholds.
- Statistical Process Control (SPC) Chart Calculator: X̄-R and individuals (I-MR) control limits with out-of-control signals.
- Outlier Detector (Z-Score, IQR, Modified Z): Flag outliers by Z-score, interquartile range fences and the robust modified Z-score.