v5.0 · MLB
Facts. Not Feelings.
Transparent sports analytics. Honest modeling, rigorous testing, public learning. We track every pick against the closing line and publish the math behind it.
3–8
High-conviction plays per day
Some days zero. That's a feature.
200+
Bet samples before a market ships
Plus +1% ROI and Brier <0.246
100%
CLV-tracked
Every pick snapped at close
Conviction tiers
Every play is tagged with a tier and a Kelly unit count. Tier tells you how confident the model is. Units tell you how much to bet. The pairing is the whole point.
≥ 4%·2u+
Rare. The model has strong, well-calibrated edge against the close.
3–4%·1.5–2u
Real edge after vig and variance. Bread-and-butter plays.
2–3%·1u
Edge is there, position size reflects it.
1–2%·0.5u
Informational. We're noting the lean but it's below the action threshold.
< 1%·Pass
Market is efficient or model is unsure. We don't bet.
Core value
Process > Picks
Every market is gated by a rolling backtest: 200+ bets, ≥+1% ROI, Brier under 0.246. If it can't clear the math, it stays off the card. We remove markets just as readily as we add them.
Core value
CLV-first
Long-run profitability tracks closing-line value, not W/L. We snapshot every pick at close and publish 30-day rolling CLV alongside hit rate and ROI. Beating the close is the actual edge.
Core value
Radical transparency
Brier scores, ROI per market, model versions, calibration, and limitations are all public. We show our work — including the markets where we currently can't beat the book.
Today's plays. Or none, if the math says pass.
New card published every morning by 11:00 AM ET. Each play priced and tier-graded against the live market.