We grade ourselves on the closing line.
CLV — closing line value — is the overfit-resistant test of a real edge: not wins, not luck. Across 4 seasons tested out of sample, the model beat the closing line (a positive CLV edge, not a claim of winning every year) in every single one — and the pooled return across all of them is positive.
or +22.8% on amount risked at our ½-Kelly staking — the sizing the picks use (317 staked, the rest Pass).
Closing-line value by tested season
CLV is the model's edge against Vegas's closing number — the metric that survives out-of-sample. It was positive in every tested season (2022–2025), so the running total above only climbs. The pooled vig-adjusted ROI (+11.1%) is bet-weighted across all 4 of them.
Live 2026
Graded picks — the live ledger
The sections above are sealed out-of-sample backtest evidence — history the model was tested against, never traded live. This section is different: the live graded ledger. Picks are members-only before kickoff; every pick becomes public here after it settles, fully graded — the frozen bet-time price, the close, CLV vs close, and the result. Win/loss is not profitability — CLV vs the close is the durable signal.
Picks are members-only before kickoff. Every pick becomes public here after it settles, fully graded — the matchup, the pick, the frozen bet-time price, the close, CLV vs close, and the W/L/VOID result, with week records and running season totals. Nothing is hidden and nothing is retro-edited: the ledger is append-only, and a re-settle is always visibly marked.
Membership = picks before kickoff — the graded history stays public for everyone.
Win/loss is not profitability — CLV vs the close is the durable signal, and every settled pick shows both.