Forecast Bias
The systematic tendency of a forecast to be too high or too low.
Formula
Bias = Sum of (Forecast - Actual) / Sum of actual demand
Worked examples
- 01Six months: forecast 6,600, actual 6,020. Bias = (6,600 - 6,020) / 6,020 = 9.6%; the forecast is too high.
- 02Monthly errors of +40, -30, +55, -20, +10, -45 units on 2,400 units actual: bias = 10 / 2,400 = 0.4%, no systematic bias.
- 03Process pumps: forecast 1,280, actual 1,167 over a quarter = 9.7% too high, which explains growing stock.
Pitfalls
- Accuracy can look fine while a steady bias builds stock or shortages.
- Short periods are noisy; look at several months.