Time and trends
Forecast accuracy
- Time and trends
Definition
How closely forecasts matched actual outcomes over a defined period. Different accuracy measures have different limits.
For your role
Choose your role to see what this term means in your work.
Why it matters
Forecasts drive stock, staffing and cash plans. Measuring accuracy, and whether errors lean one way, shows whether the forecast can be trusted and where to improve it.
Where it is used
Financial planning, demand planning and sales forecasting. Used by FP&A analysts, demand planners, supply chain managers and data scientists.
How to calculate it
Absolute percentage error (APE) = |actual - forecast| ÷ |actual| × 100 (only when actual is non-zero)
MAPE = mean of APE over the evaluated observations
WAPE = sum of |actual - forecast| ÷ sum of |actual| × 100 (only when the denominator is non-zero)
Some teams use 100 - MAPE or 100 - WAPE as an 'accuracy %', but that convention is not universal and can produce negative values.
Worked example
Take a simple forecast for Raket Parts B.V.: each month's revenue will equal last month's. For April to September 2025 the errors were 27.2%, 5.2%, 40.8%, 7.3%, 4.4% and 20.3%. MAPE = 17.5%, so forecast accuracy is about 82.5%. WAPE = €6,896,050 total error / €39,548,867 total actual = 17.4%.
Raket Parts B.V. is a fictional company created for Insyt.
MAPE is undefined if any actual value is zero and can be unstable when actual values are close to zero. The absolute-value denominator does not make percentage errors meaningful for every negative-valued series. For zero-heavy, intermittent or negative-valued data, consider MAE or a suitable scaled error instead. State the measure, aggregation level and forecast horizon before comparing accuracy.