01 · Data is source of truth
Every pick traces to a real line.
Live odds from multiple sportsbooks on a regular schedule. The model's view of a matchup is always reconcilable to the markets it was pulled from. No synthesized numbers. No back-tested narratives masquerading as live signal.
02 · Reasoning is published
If the work can't be shown, it doesn't ship.
On Pro and Elite, each pick exposes its factor breakdown: consensus, market depth, line movement, intelligence layers, and the confidence score the model assigned. Free sees the daily two-pick teaser without the confidence score. On book-priced moneyline picks the only probability shown is the market-implied one fixed in the pick's proof receipt. The calibration we measure ourselves on is the calibration of that market-implied probability on our settled two-way moneyline picks: the average implied probability across books, normalised to remove the vig, fixed at publish time and committed to the pick's proof receipt, never recomputed. Confidence is a ranking score and is not part of that measurement. You see the inputs. You decide what to do with them.
03 · Outcomes are uncertain
Perspective, not certainty.
No certainty theater. No guarantees. Even a pick the market prices as a clear favorite loses a real share of the time, on every tier. Every public surface is designed around that reality. Variance is described, not hidden.
04 · Trust is earned slowly
Performance stats stay gated until they're honest.
The public win-rate readout doesn't appear until enough settled picks exist to make it statistically meaningful. Until then, the Performance page says "Collecting." Patience over noise. That's the standard.