Most platforms show you their wins. This page shows everything — every prediction the engine has published, scored against what actually happened, wins and misses side by side. The receipts, not the promises, are the product.
Read it like this: when the engine says “60–70%”, the green bar shows how often those calls actually came true. The closer the pairs, the more honest the number. This chart is computed live from every resolved prediction — it cannot be curated.
A question in plain language becomes measurable claims, each bound to its own per-game record.
Dozens of instruments vote; dependence between factors is measured from real shared games — never assumed.
The day’s published slates are cryptographically fingerprinted and timestamped by an independent authority — before first pitch. No edits after the fact are possible.
Next morning, official results decide each claim. Correct or missed — the verdict is stamped and appears here, permanently.
When the data can’t support an answer, it doesn’t get one. Claims without sufficient real evidence are refused, never estimated. Past results never guarantee future ones — this ledger exists so you can judge the instrument by its record, not its promises.