Forecast Error

Forecast error is usually calculated as the actual value minus the forecast, although some use the reverse. A positive error under this definition means the forecast was too low; a negative error means it was too high.

For example, if a company forecast sales of 500 and actual sales were 460, the forecast error is minus 40, meaning sales were overestimated by 40, or 8%.

Analysing forecast errors over time helps improve forecasting. Errors that average close to zero suggest the forecast is unbiased. Errors that are consistently positive or negative indicate bias, such as always being too optimistic. Large errors indicate low accuracy.

Common measures summarise errors across many forecasts, including mean absolute error, root mean squared error and mean absolute percentage error. Each emphasises different aspects, such as penalising large errors more heavily. Forecast errors are inevitable, so businesses often plan using ranges and scenarios rather than relying on single-point forecasts.