Breakout upside and cost-controlled trade value for every skater, optimal deployment and farm pipeline for all 32 teams, and a live trade analyzer that tells you who wins the deal.
WAR (wins above replacement) is a box-score estimate of a player's total on-ice contribution — scoring, shot volume, two-way events (takeaways, blocks, hits, giveaways), on-ice differential and discipline — netted against a positional replacement level over ice time and converted to wins.
Surplus value is the headline number, in $M, and it is what a club gains by holding a contract rather than what the player produces. A club can always sign a minimum-salary body, so only pay above the minimum counts as cost, and only production above what that body would give counts as return. A contract a replacement could plausibly have signed is worth nothing however good the name on it; one the player would sooner tear up is worth a great deal. The rate converting wins into dollars is measured off the league's own payroll, not assumed. Crucially the horizon runs to the end of control, not the end of the season: a pending RFA in his last cheap year is a good player on an expiring number, which is a different thing from a good asset, because whoever acquires him inherits the raise.
Draft picks are priced from what draft slots have historically returned — mean career point shares by pick, over twelve settled classes — rather than from a table of assumed values. That curve is far steeper than the guess it replaced: first overall has returned roughly seven times what the 32nd pick did. Prospects start at their slot's value and move from there on what they have actually done since, in whatever league they are playing, converted to an NHL-equivalent rate. Optimal lines rank each roster by a production/role score; contract projections use CapWages' own model where available; the analyzer compares surplus on each side and flags who wins.
Calibration. Where the page can measure something it does, and where it cannot it says so. The dollars-per-win rate, the pick curve and the fairness band are fitted from evidence — league payroll, settled draft classes and several hundred realized trades. The exchange rate between picks and players is the hardest of these to pin down, because the trades that quote it most cleanly are deadline rentals, where clubs pay in draft capital for a playoff run rather than for surplus; when the fit lands outside a plausible range it falls back to a documented prior and the figure below says priors rather than fitted.
The money adds up, by construction. Wins are priced convexly — nine wins in one roster spot cannot be bought nine times at the average rate — but a convex premium left unchecked invents money that does not exist: an earlier build valued the league's skaters at 116% of the entire salary cap and put one player at $61.8M for a season against a $104M ceiling. The premiums are therefore normalised so that the league's total valuation equals the league's total payroll, which is the one identity a market model has to satisfy: in aggregate, what clubs pay is what wins cost. The premium then does only what it should — move value from the replaceable to the scarce — rather than manufacture it.
What this model is bad at. Three things, stated plainly, because a model that lists no weaknesses is not being honest about being a model. Defensive defencemen are its known blind spot: everything here is built from the box score and the on-ice differential, and a shutdown defenceman's value lives mostly in events neither one records. The defensive half of the differential is weighted above the offensive half for defencemen to compensate, but a Jaccob Slavin will still grade below where a scout would put him. Goaltending is a save-percentage estimate against a fitted replacement level, with no adjustment for shot quality — it will over-reward a goalie behind a good defence. Prospects are graded on production converted to an NHL-equivalent rate, which is a public estimate rather than a measurement, and least reliable for the smallest leagues. None of this is advice, and none of it is a licensed feed.
Data. Stats and rosters from the NHL API (current teams, so offseason moves are reflected). Cap hits, UFA/RFA status and next-contract projections live from CapWages (contracts expiring 2026–2029; players beyond that show cap n/a). Draft classes and the pick curve from Hockey-Reference; prospect production from HockeyDB; trade history from ESPN. Headshots and club marks are loaded from the NHL's media CDN. Best-effort model estimates from public data, not advice — for a licensed contract feed use PuckPedia.