Best closing-line value right now: WNBA Props +1.59 · weakest: MLB First 5 -0.23. CLV above zero means you beat the number the line closed at — the earliest sign an edge is real, long before the win/loss record settles.

Each dot: when the model gave picks about this chance (across the bottom), how often they actually hit (up the side), over 555 graded picks. Dots on the dashed line mean the model’s percentages mean exactly what they say — that’s what makes them trustworthy, even where the model still can’t beat the closing price.

11 plays on the board today. Model’s biggest lean: Portland Fire SPREAD 1.5 +12.2 (Phoenix Mercury @ Portland Fire). A bigger edge just means the model disagrees more with the book — interesting, but only real value if it beats the close. Nothing here is staked.

Avg CLV +0.04 across 30 graded — holding above zero so far — the sign to watch for. Not staked: the NFL model loses to the sharp close in backtest, so it earns real paper stakes only if this CLV holds above zero (378 still pending). The grey line under each pick is the why — weather, a red QB OUT flag, or the defense the player faces.

Avg CLV +0.33 across 252 graded — holding above zero so far — the sign to watch for. Not staked: the College FB model loses to the sharp close in backtest, so it earns real paper stakes only if this CLV holds above zero (238 still pending).

Opponent-adjusted team-yardage projections. No odds feed exists for these markets, so this is an input to compare against your own book — not a tracked edge, and not proven to beat a line. Leakage-safe accuracy (MAE vs actual): rush ~55 yds, pass ~68 yds — better than a naive guess, but single-game yardage is noisy. "pass/rec" = team passing yards, which equals team receiving yards.