Devon Levi#27
Goalie with Buffalo, 2022–23 to 2024–25
Each line above is that metric's own value across the last 10 appearances, not smoothed, plotted against the season baseline (the goalie's own save rate for SV%, zero for GSAx, the goalie's own quality-start rate for Quality Start %); the thin vertical rule splits off the last 5, which is what that column reports.
| Game | SV% | Saves | xGA | GSAx | Rest |
|---|---|---|---|---|---|
| Jan 20 at SEA | .848 | 28 / 33 | 3.4 | -1.6 | 36d |
| Dec 15 at TOR | .900 | 36 / 40 | 4.8 | +0.8 | 29d |
| Nov 16 at PHI | .871 | 27 / 31 | 3.4 | -0.6 | 2d |
| Nov 14 vs STL | .889 | 24 / 27 | 2.2 | -0.8 | 3d |
| Nov 11 vs MTL | .500 | 2 / 4 | 1.1 | -0.9 | 10d |
| Nov 1 vs NYI | .889 | 32 / 36 | 2.9 | -1.1 | 15d |
| Oct 17 at CBJ | .760 | 19 / 25 | 1.8 | -4.2 | 5d |
| Oct 12 vs FLA | .920 | 23 / 25 | 1.6 | -0.4 | 7d |
| Oct 5 at NJD | .919 | 34 / 37 | 3.2 | +0.2 | – |
A quality start allows no more goals than that night's shots faced predicted, a really bad one allows 2 or more goals above that prediction. Relief appearances are listed here but get neither verdict, so they sit outside the Quality Start % above. Descriptive only, small samples swing.
Career With Buffalo
All-time is the rank among the 63 goalies with a regular-season game for Buffalo since 1970–71. W-L ranks by wins; GAA and SV% count only careers of at least 25 games (34). Ties share a rank.
Season by Season
GSAx by Situation
Levi's shots faced in 2024–25, split by strength state and by shot danger. GSAx is the total expected goals of those shots minus the goals allowed: a positive number means stopping more than an average goalie would on that workload. The smaller number beside it is one standard deviation of that GSAx, given the shots faced in that split treated as independent coin flips at their own odds. A split whose GSAx sits inside its own ± band is not distinguishable from an average goalie on that workload, which is common for the smaller penalty-kill and high-danger samples even across a full season.
| Split | Shots faced | xGA | GA | GSAx |
|---|---|---|---|---|
| All | 352 | 24.2 | 33 | -8.8±4.5 |
| 5v5 | 274 | 16.6 | 25 | -8.4±3.8 |
| Power play (Buffalo up a skater) | 8 | 0.7 | 0 | +0.7±0.8 |
| Penalty kill (Buffalo down a skater) | 70 | 7.0 | 8 | -1.0±2.3 |
| High danger | 62 | 12.7 | 21 | -8.3±3.0 |
| Medium danger | 100 | 7.9 | 8 | -0.1±2.7 |
| Low danger | 190 | 3.6 | 4 | -0.4±1.9 |
Shots faced counts every unblocked attempt the model scores, shots that missed the net included, so it runs higher than saves-plus-goals. High danger is an attempt worth 0.12 xG or more, the same rule every high-danger count on the site uses; medium and low split the rest by expected-goal value. Empty-net and no-coordinate shots count in the totals but sit outside the strength and danger splits, so the rows need not add up exactly.
GSAx Through the Season
The running season total, ending on the GSAx the splits table below shows for all situations. Ticks under the axis mark quality starts (green) and really bad ones (red).
Save Map
Every unblocked shot Levi faced in 2024–25, to scale, all attacking the same net. Dot size is the shot's expected-goal value; goals against are filled red. Narrow it down and the numbers below follow the slice.