Simulation Lab
Tune any FMCG formula live — forecast accuracy, EOQ, fill rate, OTIF, OEE, GMROI, working-capital cycle. Defaults reflect a Masan-style sauce/noodle/coffee operation.
FMCG Simulation Lab — 36 formulas, 8 domains
Every metric a Masan-style FMCG planner uses, in one page. Each card now shows a sensitivity sweep — drag the slider or pick a different input to vary, and watch the metric trace across the band-coloured healthy / watch / action zones.
WAPE (Weighted Absolute Percentage Error)
Volume-weighted error. Penalises misses on high-volume SKUs more than low-volume — the right accuracy metric for FMCG where the long tail dominates.
FA = 87.2%
WAPE = Σ|Actual − Forecast| / Σ|Actual| × 100%
FMCG WAPE target: ≤ 25% category, ≤ 35% SKU/store
MAPE (Mean Absolute Percentage Error)
Equal-weight average APE. Easy to communicate but biased on small denominators (zero-actual periods inflate it).
MAPE = mean(|Actual − Forecast| / Actual) × 100%
FMCG MAPE: ≤ 20% aggregate, ≤ 40% SKU-level
Forecast Bias
Persistent over- or under-forecasting. Bias drives safety stock size — a +10% bias means SS is too low, a −10% bias means you're overstocking.
Bias = Σ(Forecast − Actual) / Σ(Actual) × 100%
Healthy: |bias| ≤ 5%. > 10% = retune model
Tracking Signal
Cumulative bias divided by mean absolute deviation. |TS| > 4 = model is drifting and needs intervention.
TS = Σ(Forecast − Actual) / MAD
OK: |TS| ≤ 4 · Drift: |TS| > 4 · Out-of-control: |TS| > 6