Free-Agent Class Valuation

The very best players read low — by design. Generational stars almost never reach free agency, so the market gives the model only a handful of examples of their pay (and those are re-signings, not open-market deals). Rather than invent a premium from that, the model prices the top few on measurable on-ice value and shows the market comp beside it. Their rating is correct — the gap is the re-signing, term, and marquee premium we deliberately don't extrapolate, not a flaw in the player. We tested closing it (twice); the data won't support it, so we show both rather than guess.

Projections skew high for veteran UFAs. Free agency selects for players past their peak — the aging curve describes survivors, not the average career. We are measuring this live and will calibrate. Rows for players 32 or older carry a VET flag.

The RFA discount is a league average; it understates star RFAs, who sign near their open-market worth. For high-end RFAs, treat the status-adjusted headline as a floor — the open-market figure (shown beneath it) is the more realistic anchor.

Defenseman value uses a deployment-adjusted impact model (RAPM) that isolates each player from his linemates, competition, and zone starts. It covers 5-on-5 play and venue only — special teams and shot-blocking beyond their effect on 5-on-5 expected goals are not captured.

Confidence reflects sample size and is capped at "med" — a projection is never high-confidence. Goalies are always flagged "low": goalie value repeats far less reliably year to year and the free-agent goalie pool is small, so those projections are the least certain on the board and can swing hard between updates. The low–high range is the model's sensitivity band (status-discounted and floored at the league minimum), not a guarantee.