Motion-blurred scrum of skaters in front of the net

Power rankings · 2026-27 · updated Oct 10, 2026

Who's actually good.

Every team rated in goals per game against an average team. Records lie early; this is built from how teams play, their special teams, and how good their lineup has been, in that order as the season fills in.

69 games played league-wide · +1.00 = a goal a game better than average

The board

#	Team	Record	Rating	Track record	Goals	5v5 shots	PP G	PK GA
1	Carolina HurricanesCAR	3-1-1	+0.99	80%	18-14	60.8%	7	3
2	Vegas Golden KnightsVGK	4-1-0	+0.75	80%	21-13	58.2%	6	2
3	Tampa Bay LightningTBL	3-1-0	+0.65	83%	11-9	57.0%	1	2
4	Colorado AvalancheCOL	3-1-0	+0.46	83%	23-11	53.0%	4	1
5	New Jersey DevilsNJD	1-2-0	+0.41	87%	6-13	61.0%	1	4
6	Buffalo SabresBUF	2-2-0	+0.37	83%	10-15	60.3%	0	4
7	Utah MammothUTA	3-2-0	+0.35	80%	18-14	58.1%	8	4
8	Edmonton OilersEDM	3-0-1	+0.33	83%	22-16	53.1%	4	5
9	Florida PanthersFLA	2-0-2	+0.28	83%	8-8	60.3%	2	4
10	Minnesota WildMIN	2-1-1	+0.24	83%	11-8	48.8%	5	1
11	Pittsburgh PenguinsPIT	2-2-1	+0.21	80%	20-16	49.4%	4	3
12	Ottawa SenatorsOTT	3-1-0	+0.16	83%	12-9	46.3%	2	3
13	Dallas StarsDAL	2-2-0	+0.16	83%	10-6	53.4%	1	2
14	Columbus Blue JacketsCBJ	2-1-0	+0.11	87%	10-9	51.1%	2	3
15	Washington CapitalsWSH	3-1-0	+0.01	83%	14-9	49.8%	2	2
16	Montreal CanadiensMTL	1-2-1	-0.02	83%	14-19	51.0%	4	5
17	New York RangersNYR	4-2-0	-0.04	77%	17-11	45.7%	4	2
18	St. Louis BluesSTL	1-2-1	-0.15	83%	9-13	49.6%	1	1
19	Los Angeles KingsLAK	0-2-1	-0.20	87%	9-15	52.2%	3	5
20	Anaheim DucksANA	3-1-0	-0.23	83%	14-11	48.9%	5	3
21	Winnipeg JetsWPG	3-1-1	-0.31	80%	13-15	45.7%	4	1
22	Nashville PredatorsNSH	2-1-1	-0.37	83%	12-11	41.6%	5	2
23	Toronto Maple LeafsTOR	2-2-1	-0.38	80%	14-15	48.9%	3	2
24	Seattle KrakenSEA	3-2-0	-0.40	80%	21-14	44.0%	4	4
25	Boston BruinsBOS	3-2-0	-0.40	80%	15-12	49.7%	3	3
26	Detroit Red WingsDET	1-3-0	-0.44	83%	10-14	58.9%	3	6
27	Philadelphia FlyersPHI	0-3-2	-0.56	80%	6-19	35.4%	2	5
28	New York IslandersNYI	2-2-0	-0.58	83%	13-8	45.9%	3	3
29	San Jose SharksSJS	3-1-0	-0.75	83%	12-14	42.1%	2	3
30	Calgary FlamesCGY	0-4-0	-1.00	83%	6-23	51.9%	2	1
31	Chicago BlackhawksCHI	1-4-0	-1.09	80%	10-21	42.2%	2	5
32	Vancouver CanucksVAN	2-3-0	-1.47	80%	21-25	40.3%	3	8

Rating = goals per game vs an average team. The bar splits it into its 5v5 part and its special-teams part (the second colour; hover a bar for each number). Track record = how much of the 5v5 part still rests on how good this lineup was last season; it fades as games are played. Goals include overtime, not shootouts.

Points % minus what the rating expects

Record vs rating

Winning more than their play

San Jose SharksSJS3-1-0 · #29 · +56
Winnipeg JetsWPG3-1-1 · #21 · +36
Nashville PredatorsNSH2-1-1 · #22 · +30
Anaheim DucksANA3-1-0 · #20 · +28
Edmonton OilersEDM3-0-1 · #8 · +27

Playing better than their record

Calgary FlamesCGY0-4-0 · #30 · -40
Detroit Red WingsDET1-3-0 · #26 · -39
Los Angeles KingsLAK0-2-1 · #19 · -36
New Jersey DevilsNJD1-2-0 · #5 · -30
Buffalo SabresBUF2-2-0 · #6 · -23

The number is percentage points of points %: a team at +30 has 30 points more per 100 possible than its rating would expect. Early in a season most of that gap is bounces.

Our model, our data

How it's built

Three parts, one number

Rating = (5v5 share blended with the track record − 50%) × goals per share point + special-teams goals per game × its weight. Everything is per game, so a team that has played six games gets no edge over one that has played three. Model team-power-v2, rebuilt every morning from the NHL's play-by-play.

5v5 play: Each team's share of the expected goals at five on five, adjusted for score: who has the puck and who gets the better chances. Converted to goals per game (one point of share ≈ 0.13 goals a game).
Track record: Early games are mostly noise, so the 5v5 part starts from how good this season's lineup was last season, player by player, weighted by the minutes they're playing now and adjusted for age. It counts as 20 games of evidence: about 82% of the 5v5 part today, half by game 20, very little by spring.
Special teams: Power-play goals minus penalty-kill goals against, per game, trusted more as games pile up (weight = games ÷ (games + 35)).
Left out: Hot shooting and hot goaltending. In our tests neither predicted the rest of a season, so they don't move the rating; they show up in the record-vs-rating gap instead.

Tested on 2023-24 to 2025-26, 4 checkpoints

Does it work?

After	This rating	Points %
5 games	0.64	0.27
10 games	0.68	0.35
20 games	0.69	0.45
41 games	0.69	0.64

Correlation with each team's goal differential over the rest of that season, tested on seasons the model wasn't fitted to. Higher is better. Early on the rating predicts about twice as well as the standings; by mid-season they meet.