ToolJoltTools

MAPE & sMAPE Calculator

Percentage forecast error and its symmetric variant — the demand/sales forecasting accuracy metrics, with their pitfalls.

MAPE (%)
sMAPE (%)
WAPE (weighted) (%)

MAPE is intuitive but explodes near zero actuals and asymmetrically punishes over-forecasting. sMAPE bounds it (0–200%) and is symmetric. WAPE (a.k.a. MAD/Mean) is the robust choice for intermittent demand with zeros — increasingly the supply-chain default.

Formula

MAPE = mean|A−F|/|A| · sMAPE = mean|F−A|/((|A|+|F|)/2) · WAPE = Σ|A−F| / Σ|A| (robust to small actuals)
References: Hyndman & Koehler (2006), Another look at measures of forecast accuracy; Makridakis (1993), Accuracy measures: theoretical and practical concerns

About MAPE & sMAPE Calculator

Forecasters love percentage errors because '12% off' is instantly meaningful to stakeholders — but the popular MAPE hides three traps: it's undefined when actuals are zero, it explodes for small actuals, and it punishes over-forecasting more than under-forecasting. This calculator computes MAPE alongside its symmetric cousin sMAPE (bounded and fairer to both error directions) and WAPE (volume-weighted, robust to zeros and intermittent demand). Comparing the three on your forecast reveals whether percentage thinking even applies to your data — and which metric you should actually report.

How to use MAPE & sMAPE Calculator

  1. 1Enter your values into MAPE & sMAPE Calculator — sensible, domain-typical defaults are pre-filled so you see a real result immediately.
  2. 2The result recomputes live using the formula shown on the page; there is no button to press.
  3. 3Adjust any input to compare scenarios, then read the worked example to see the substituted numbers.

Why use MAPE & sMAPE Calculator?

  • Computes MAPE & sMAPE instantly in your browser — no sign-up, no upload, no server round-trip.
  • 100% free and unlimited, with the exact formula shown: MAPE = mean|A−F|/|A|.
  • Runs entirely client-side, so every value you enter stays private on your device.
  • Live recompute as you type, with a worked example and authoritative references for trust.

Frequently asked questions

Why does MAPE break on data with zeros?+

It divides by the actual value, so any zero actual produces division by zero (undefined), and near-zero actuals produce enormous percentage errors that dominate the average. For intermittent demand — products that sell zero units some periods — MAPE is unusable; WAPE, which divides total absolute error by total actuals, is the standard fix.

What's the difference between MAPE and sMAPE?+

sMAPE divides the error by the average of forecast and actual rather than by the actual alone, making it symmetric (over- and under-forecasting are penalized equally) and bounded between 0% and 200%. It's more robust but less intuitive — a 100% sMAPE doesn't mean 'off by 100%' the way people expect.

Why is MAPE biased toward under-forecasting?+

Because the error is divided by the actual: a forecast 50% too high on a value of 100 gives 50% error, but a forecast that can be at most 100% too low caps the downside differently. Optimizing for MAPE quietly nudges models to under-forecast. sMAPE and WAPE don't share this asymmetry.

Which forecast metric should supply-chain teams use?+

WAPE is increasingly the default: it's interpretable as a percentage, handles zeros gracefully, and weights errors by volume so high-selling SKUs matter more than slow movers. Report MAPE for stakeholder communication if data has no zeros, but track WAPE (or MASE, scaled against a naive forecast) for honest model comparison.

Related tools

Related ML & AI tools

Sponsored