Proof
Model Health.The model, judged in public.
Most products tell you their model is good. Ours publishes the measurement that would prove it wrong: the Brier score — the average squared gap between what we announced and what happened — league by league, market by market, bad spots included.
A plain-words scale
0 would be perfect; 0.25 is a coin flip. Every screen that shows a Brier score explains it in one sentence, without jargon — a measurement nobody can read is a decoration.
Refreshed weekly, tested honestly
Models are refitted every week and always scored on seasons they never saw. A new model climbs through shadow, canary and partial stages — nothing is promoted the day it is born.
The measurement governs the product
When a league runs ten points below what was announced over twenty judged analyses, it is flagged on every fixture and loses the strong signal. Disproved markets are withdrawn. No enthusiasm can overrule the record.
No verdict on thin evidence
Under twenty judged analyses, a gap is sample noise: the product refuses to call a league good or fragile, and says how many it has.
The live measurement is in the app.Open it and judge.
Next: the verified record · the method