Research
The evidence room: every claim the product makes is graded here, in the open — losses included.
Track record
Every logged selection graded against the closing line — CLV, ROI, receipts. The number the whole product answers to.
Signal explorer
Attribution for every measured signal (AI, market, referee, availability): win-rate lift vs the engine baseline, sample sizes, keep/monitor/drop verdicts.
Backtest lab
Held-out walk-forward validation — parameters frozen on training seasons, replayed on untouched later seasons. A historical simulation, deliberately kept off the live ledger.
Methodology
How the model, the market pricing and the grading actually work, end to end.
Engine score
What the 0–100 process score is — and why it is never a win probability.
Data landscape
Every provider we use, what each is allowed to influence, and the budgets that govern them.