Every metric name opens its definition; the full list lives in Metrics.
Traditional statistics cover all situations. Supporting five-on-five analyses use completed games, raw adjustment and exposure from xG model v4. The two samples can differ.
An em dash means unavailable, not zero. A partial season remains partial. Where a panel reports evidence coverage, its rates use only the covered games. History never substitutes a different model, and a missing appearance interrupts a rolling model window.
Shot maps use the same shot model as Game Center and show which model version scored recent shots beside the rink. Older seasons can use earlier model versions. High danger means a model scoring probability of at least 0.20 on an unblocked attempt; for goalies, medium danger is 0.08 to 0.20 and low is below 0.08.
Special-team buckets include the wider recorded man-advantage and man-down states, read from the player’s own team; for a goalie, penalty kill means facing a power play. Their ranks need 100 recorded minutes and at least 30 qualified peers; below 100 minutes, penalty-kill percentiles are provisional. Playoff values have no ranks or percentiles. Team share is player ice time divided by team ice time in that situation, over the games the player dressed. Zone starts describe faceoff-started five-on-five shifts and exclude on-the-fly starts.
Penalties count events, including one for a double minor; bench and served-by roles are excluded, and victimless infractions have no drawn role. EPV added sums changes in the probability that the possessing team scores next; it is not standings points or an isolated player grade. Rolling windows never cross seasons or season types, and a missed game is not a zero-performance appearance. Career rates use summed counts and exposure, never an average of season rates; a missing contributing season withholds its career metric, and Performance begins in 2023–24.
Ice time is per game played; minutes outside the three recorded scopes appear as Other. 6 all-situations games.
.886 save percentage over the last 6 games (117 saves on 132 shots)
.886 save percentage over the last 6 games (117 saves on 132 shots)
6 GP · Last start: W 4–3 vs SEA (24 saves on 27 shots)
6 GP · Last start: W 4–3 vs SEA (24 saves on 27 shots)
Qualified · 2026–27 goalies
Provisional comparisons · Small sample.
Goals saved above expected
-2.66
Five-on-five
Expected goals against
9.34
12 allowed at five-on-five
Save percentage
.880
100 five-on-five shots
Goals saved above expected
-0.93
All situations
Saves by shot danger · five-on-five
Danger
Shots
Share
Goals against
Save %
High danger
23
23%
7
.696
Medium danger
30
30%
3
.900
Low danger
47
47%
2
.957
How this is calculated
Scoring
Season grade is the qualified peer percentile on the same letter bands as Game Grade — A+ ≥88, A ≥78, A− ≥68, B+ ≥58, B ≥50, B− ≥42, C+ ≥35, C ≥28, C− ≥22, D ≥15, F below 15 — and needs qualification plus at least 30 eligible peers. The letter describes this season’s performance, not permanent ability.
Comparisons
Rank and component percentiles compare this season’s qualified goalies on unrounded rates over every checked appearance; Game Grade compares one appearance with frozen historical games.
Coverage
Qualification: 2 covered appearances, 64.0 verified 5v5 shots, and 90% appearance and workload coverage. Minimums advance with league progress; valid low-sample results stay visible.
Zero-shot relief appearances do not count toward qualification until eligible five-on-five ice time is verified.
Coverage: complete. Updated .
Method
Stopping uses the independent verified five-on-five shot-on-goal model. Danger and rebound measures have separate gates and are not added to this score.
-2.00 total goals saved above expected.
Legacy rating
5v5 legacy season score: 13.8 / 100. The fixed exposure baseline is 50; this is separate from the season peer grade.
Recent form
Trailing ten appearances. Shaded games have fewer than ten in the window.
Trailing 10 games · Five-on-five GSAx / 100 shots
-2.66 GSAx / 100 SOG
Oct 15, 2026vs SEA
Trailing 6 · Five-on-five GSAx / 100 shots
-2.66
6 appearances in window · partial window6 with five-on-five model coverage
Stopping and rebounds measure results. Stoppages describe style; shot difficulty and workload describe what the goalie faced. A larger shape is not a better season.
Among G peers · 57 qualified players · Descriptive profile · performance, style and context
Player value: 0.55 Peer range: -0.89 – 1.32 goals saved per 100 SOG
MeanAndrei Vasilevskiy
Lower Higher (better)
Player 0.55 · mean 0.20. Higher raw values rank higher.
Adjusted goals saved above expected per 100 verified five-on-five shots on goal. Higher is better.
How to read these comparisons
Each axis ranks the player against qualified peers. The dashed reference is the percentile of their arithmetic mean, which need not be the 50th percentile. Ties share a midrank.
This shape combines performance, playing style and the difficulty and volume of shots faced. Farther from the center means more of that named quantity. A larger shape is not an overall grade. Post-save stoppages are a recorded event proxy, not goalie-attributed freezes.
Goalies need 15 covered appearances, 500 verified five-on-five shots on goal and 90% shot and appearance coverage. Rebound suppression needs 600 eligible saves across 15 appearances. Post-save stoppages need 600 observable saves across 15 appearances, at least 90% horizon coverage and 98% reason coverage. At least 30 goalies must qualify per axis. Playoff ranks are withheld. A reviewed axis whose own evidence fails is withheld and shown with its reason rather than ranked. These workload rules do not express statistical confidence.
Position uses the NHL listed primary position available now; it does not establish historical position or actual deployment. Broad groups are forwards, defence and goalies. This version uses the current model’s covered games only.
Season vs reference
All situations
Metric2026–27ReferenceDiff
.886.911−.025
per 60 minutes2.532.31+0.22
per game22.0025.57-3.57
Reference: 2025–26 · 58 games.
Recent-history profile
Loading recent-history profile…
Goalie history radar
Stopping and rebound results across recent NHL seasons, compared with a fixed goalie reference. Includes previous seasons; separate from the selected-season standing above.
Through Oct 15, 2026. Radar and percentiles use the same checked history; season ranking qualification does not hide this profile.
Bands show position percentiles; axis labels show current rolling values.
Dashed outline: average goalie in the fixed reference. Outside = above average; inside = below average. All axes point outward for stronger results.
Player value: 0.74 Peer range: -0.60 – 1.49 residual per 100
MeanAndrei Vasilevskiy
Lower Higher (better)
Player 0.74 · mean 0.49. Higher raw values rank higher.
Rolling profile against a fixed reference group
Fixed rolling-profile reference · 52 players. Values use the same history and smoothing as the radar.
History, sample sizes & calculation
Higher radar percentiles mean stronger recorded results against the fixed reference, including older seasons. These are separate from this season’s league percentile. Each player has a different amount of covered history. Weighted minutes and attempts give recent games more influence. “Prior” is the share contributed by a frozen league baseline: it is higher when a player has less evidence, and differs by metric because each uses its own smoothing strength. It is not a confidence score.
Overall stopping · residual per 100
1170.1 weighted verified SOG · 46% prior
Mid-danger stopping · residual per 100
420.0 weighted verified SOG · 37% prior
Low-danger stopping · residual per 100
656.9 weighted verified SOG · 75% prior
Rebound suppression · residual per 100
1058.2 weighted eligible saves · 79% prior
Verified 5v5 stopping and rebound profile. High-danger stopping is withheld; these overlapping axes are not added together.
History through Oct 15, 2026. Recent games count more than older games. History carries across seasons and stays unchanged while inactive.
