Shelly Β· quality Β· v Β· synthetic data

Can the numbers be trusted?

An analyst's work is only as good as its checks. This page shows them: how every forecasting model scores and which one is in charge, the trading days that don't look right and the likely reason, what was checked in the raw data before any model ran, and the automated tests that guard every release.

Forecast league table

Error on weeks the models never saw Β· lower is better

Best model by category

Anomaly radar

Data checks Β· before any model ran

What was found in the raw store data, and what was done about it

Build health

What guards every release