Skip to content

Insights

Where the numbers say something a scoring table does not: postseason production, finishing and goaltending against the xG model, and scoring on the power play.

Season
2026-272025-262024-252023-242022-232021-22
Hottest finisherEvan Bouchard+7.2goals above the model · 11 where the model expects 3.8Coldest finisherZach Benson−6.1goals below the model · 2 where the model expects 8.1Best goaltendingJesper Wallstedt+16.9goals saved beyond the model · 686 attempts faced

Playoff scoring, 2026-27

Points in the postseason

The games every other board on this site leaves out, ranked on the total with games played on the row.

No playoff games for this season yet

A board appears here once the season has games to rank.

Why the total, not points per game — a per-game sort crowns a two-point sweep
A per-game sort with no minimum puts a two-point sweep above a twenty-nine-point run, and games played vary more than fivefold across a field decided by elimination. So the board ranks the total, and the games are on every row.

Finishing against the model, 2026-27

How these boards are built — regular season to date, measured against the xG model
Expected goals come from the xG model over open-play unblocked attempts in the regular season so far. Each figure is the skater's goals minus the model's expected goals on the same attempts, not measured against the league average: the league-wide level of the model for this season is read once, after the season ends, so this season's boards do not add to zero and are not compared one-to-one with a finished season's. Boards sort on the total rather than the per-attempt rate, because a rate board with no minimum is a list of players with one attempt.

Above the model

Skaters who scored the most goals beyond what the model expected from the same attempts.

#
Skater
G
1
Evan BouchardD
20
87
11
3.8
+7.2
+0.082
2
Cutter GauthierL
17
117
16
9.5
+6.5
+0.055
3
Leon DraisaitlC
20
114
18
11.6
+6.4
+0.057
4
Artturi LehkonenL
19
58
13
7.5
+5.5
+0.095
5
Roman KantserovR
17
47
9
3.9
+5.1
+0.109
6
Jimmy SnuggerudR
18
64
10
5.4
+4.6
+0.071
7
Matvei MichkovR
18
47
10
5.7
+4.3
+0.091
8
Kyle ConnorL
19
119
15
10.8
+4.2
+0.036
9
Logan CooleyC
16
57
11
6.8
+4.2
+0.073
10
Steven StamkosC
18
73
11
7.0
+4.0
+0.054

Below the model

The same column sorted the other way.

#
Skater
G
1
Zach BensonL
17
73
2
8.1
−6.1
−0.083
2
Joel Eriksson EkC
20
96
6
11.8
−5.8
−0.060
3
Wyatt JohnstonC
15
69
3
8.4
−5.4
−0.079
4
Gabriel VilardiC
18
55
1
5.8
−4.8
−0.087
5
Juraj SlafkovskýL
17
64
2
6.1
−4.1
−0.064
6
Matt BoldyL
20
123
7
11.1
−4.1
−0.033
7
Timo MeierR
19
102
3
7.1
−4.1
−0.040
8
Alex LaferriereR
15
79
3
7.0
−4.0
−0.051
9
Mavrik BourqueC
15
43
0
3.9
−3.9
−0.090
10
Anders LeeL
18
65
3
6.8
−3.8
−0.059

Goaltending against the model, 2026-27

What the goalie boards count — unblocked attempts, not the box score's shots against
Attempts here are open-play and unblocked, deliberately not the box score's shots against, which counts a different set of pucks. Save percentage is not shown for the same reason: the game log already gives it over that other set. Each figure is the model’s expected goals on the attempts he faced minus the goals he allowed, season to date, not measured against a league-average goalie, so the board does not add to zero.

Above the model

Goalies who allowed the fewest goals compared with the model's expected goals on the same attempts.

#
Goalie
1
Jesper Wallstedt
17
686
33
49.9
+16.9
+2.46
2
Jacob Fowler
11
482
24
37.4
+13.4
+2.78
3
Jeremy Swayman
13
535
29
40.7
+11.7
+2.19
4
Pyotr Kochetkov
12
389
20
29.1
+9.1
+2.34
5
Juuse Saros
13
604
32
40.6
+8.6
+1.42
6
Linus Ullmark
8
281
10
18.4
+8.4
+2.99
7
Ilya Sorokin
12
516
31
38.2
+7.2
+1.40
8
Daniil Tarasov
8
288
15
21.9
+6.9
+2.38
9
Joel Hofer
13
563
30
36.8
+6.8
+1.21
10
Jake Oettinger
13
434
26
32.8
+6.8
+1.57

Below the model

The same column sorted the other way; a goalie on a heavily outshot team can sit here on volume alone, which is why the attempts faced are on the row.

#
Goalie
1
Kevin Lankinen
9
325
43
28.1
−14.9
−4.60
2
Dustin Wolf
10
419
36
28.8
−7.2
−1.72
3
Jakub Dobes
7
269
28
21.5
−6.5
−2.40
4
Karel Vejmelka
13
475
40
33.7
−6.3
−1.32
5
Scott Wedgewood
5
190
18
12.3
−5.7
−2.99
6
Sebastian Cossa
8
265
24
18.4
−5.6
−2.12
7
Jordan Binnington
5
209
20
14.4
−5.6
−2.68
8
Arvid Soderblom
5
193
21
15.6
−5.4
−2.79
9
Jake Allen
10
388
32
26.8
−5.2
−1.35
10
Darcy Kuemper
10
394
33
28.4
−4.6
−1.18

Power-play points, 2026-27

With the man advantage

Points recorded while the skater's side had the extra man.

#
Skater
G
A
PTS
2
Leo CarlssonC
2
15
17
8
6
Seth JonesD
1
12
13
16
10
Logan CooleyC
4
8
12
4
Shots here — shots on goal, not the attempts on the finishing boards
The shots column on this board counts shots on goal. The finishing boards count open-play unblocked attempts, a larger population, so the two are not the same number and should not be compared. A point counts as a power-play point when the on-ice strength state of the event gave the skater's side the extra man.
Why no club is shown — the club we hold is the current one, not the season's
We hold each player's CURRENT club, not the one they played for in the season a board describes. Among the top 50 finishers that would be wrong for 8 of 50 in 2025-26 and 29 of 50 in 2021-22, so the column is left off rather than shown wrong. Follow a name to the player's page for club and history. The club we hold is re-checked once a week, and every board on this page is updated once a day.

Does finishing persist?

Yes, partly — it beat both simpler rules, so far on a backtest only: no live projection has been graded yet.

The finishing boards show who out-scored the model; this grades how much of that the projection expected to carry into the next season, on a backtest and on the live record, which are not interchangeable.

Tested on 2025-26 — a season the rule never saw

a backtest, not a track record

483 of 490 projections scored · 7 unscoreable

The unscoreable ones are players who did not play 2025-26. They are counted rather than dropped — a rule that only looks good on the players who came back would otherwise look the same as one that works.

vs. assuming it all goes awaybetter

2.6% lower prediction error

A small margin is the honest outcome here, not a weak one — the rule keeps under a fifth of a hot season, so its forecast is close to this baseline by construction. A large number would be evidence of a bug.

average error 1.7642 against 1.8121 — a paired difference of −0.0479 [−0.0792, −0.0165]

vs. assuming last season repeatsbetter

18.2% lower prediction error

average error 1.7642 against 2.1574 — a paired difference of −0.3932 [−0.5155, −0.2709]

The live record

Nothing graded yet. Every projection is recorded on the day it is made, but none has come due yet — these are forecasts about next season, so the first ones can be graded once it has been played.

Unlike the backtest above, which can be re-run until it flatters, a prediction recorded before its season cannot be: the data it will be scored against does not exist yet.