NHL Analytics
NHL Stats
Components — each in wins above a replacement-level player EVO Even-strength offense EVD Even-strength defense PP Power play PK Penalty kill PEN Penalties drawn/taken SHOOT Shooting WAR Total (sum of components)
View
| # | Player | Pos | GP | WAR | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Loading… | |||||||||||||
On-ice defense & deployment signals — 5v5, ≥300 five-on-five minutes in the season xGA/60 expected goals against while on ice REL on-ice xGA/60 minus the team's without them (negative = team defends better with them out there) PK/GP shorthanded minutes per game (deployment trust) QoC competition faced, percentile of opposing forwards' ice time CT% crunch trust — share of the team's crunch ice time; crunch is measured from coach behavior (bench-shortening): the final minutes within 1-2 goals, briefly within 3 CT xGA expected goals against per 60 in those crunch moments, weighted by each game's measured intensity DEF% defense percentile vs same-position skaters — total expected-goal suppression in goal-impact units: everything that provably travels with a player (venue-corrected, team-season effects absorbed both sides, weld-gated), carrying traveling per-attempt quality at its natural goal weight plus a capped venue-corrected puck-security term, ledger-tested so no chance is charged twice. Credit is separated between teammates with each player's penalty scaled to his own evidence, so thin-minute credits shrink inside the solve (rev 13). Each channel's weight is bound by how reliably we measure it — the per-attempt quality channel is roughly a third as reliable as attempt suppression and now carries weight to match (rev 14). The headline blends the selected window with a recency-weighted three-season prior (weights and shrink fit on next-season prediction; the memory saves breakouts and lags declines — both faces priced in rev 12, with rev 9 · rev 8 · methods · audit behind it). Certified by team-change (mover) gates, a deployment-shock gate, and outcome agreement. LS = LOW-SAMPLE: evidence below the rating floor — the card prints with a percentile-equivalent but holds no rank and no leaderboard seat. QUAL% = the in-scheme quality exhibit; PKadj = adjusted penalty-kill results, context only (PK results don't travel: mover −0.03). Single-season changes are mostly noise — treat YoY moves skeptically. Span boards default to a ≥500 min/yr filter; untick to see everyone Span pick one season, a range, or the whole era — multi-season numbers are ice-time-weighted
to Team card →
Pos
— shown | # | Player | Pos | GP | xGA/60 | REL | PK/GP | CT% | DEF% | RAW% | QUAL% |
|---|
Team card → Goaltending · regular season · source MoneyPuck
Min GP
— shown | # | Goalie | GP | GSAx | GAA | xGAA | GAA vs Exp |
|---|