WAPE vs MAPE vs MASE: which accuracy number is lying to you today?
Three forecast-accuracy metrics, three very different lies. A field guide to reading the number before you trust it.
"Our forecast accuracy is 85%." Eighty-five percent of what, measured how? The metric hiding behind that number decides whether it's a real claim or a comfortable fiction.
MAPE — the popular one that breaks on small numbers
Mean Absolute Percentage Error averages the percentage error per period. It's intuitive until demand is small or zero: divide by a near-zero actual and one SKU's error explodes to thousands of percent, dragging the average into nonsense. For intermittent demand, MAPE is actively misleading.
WAPE — the one we lead with
Weighted Absolute Percentage Error sums the absolute errors and divides by the sum of actuals. It weights by volume, so your big SKUs count more and a single small-denominator SKU can't blow up the number. It's the honest headline for a mixed catalogue — which is why it's the number on our accuracy page.
MASE — the one that knows about naïve
Mean Absolute Scaled Error compares your forecast to a naïve baseline. Below 1, you're beating naïve; above 1, a one-step-back guess would have done better. For intermittent demand, MASE is the only one of the three that stays meaningful — which is why our intermittent bench is judged by it.
The rule
No single metric is "the" accuracy. Ask which one, ask what it hides, and be suspicious of any vendor who quotes a percentage without naming the metric behind it.
Every number in SangamIQ shows its work — see our live accuracy page, or start free and put your own CSV through it in ten minutes.