Welcome back to the start of a great season of hockey! Most teams have played six to eight games. That's enough to get excited and not nearly enough to know much. Here's how we're reading it.
How the rankings work
The rating is our win model's chance that a team beats a league-average opponent, averaged over home and away games. It runs mainly on even-strength expected goals and a rating of the skaters a team dresses. Goal differential, win rate and goaltending play smaller parts.
The model only uses this season's games, apart from the player ratings. With six to eight games each, the ratings are spread far wider than real talent. Nobody's true chance against an average team is 69% or 32%. Use the order and ignore the percentages.
A .500 team at No. 1
New Jersey is 3-3 with a −6 goal differential, and the model has them first. They're getting 60.7% of even-strength expected goals, third in the league. They've also scored about 10 fewer goals than their chances predict, net of goals allowed, on 5.1% shooting. The model looks past the finishing, and a run of games decided by shooting percentage tends not to keep going one direction.
Carolina and Utah are right behind them, and neither needs much explaining. Carolina is 6-1-1 with 61% of the chances. Utah's 61.5% leads the league on ordinary shooting (7.8%) and goaltending (.925). Utah is the one that's new: they were at 52.7% last season.
Better than their record
Calgary is 1-5 with a −17 goal differential, and the model has them 20th. Their shot share is basically even, 49.8%. They're shooting 4.3% and getting an .891 save percentage, both far below a normal season.
Los Angeles at No. 13 with no wins is the same story at a smaller size: 52% of the chances, 4.7% shooting, .888 goaltending.
Worse than their record
Rangers (6-2, +13, No. 17). They have the league's best goal differential with less than half the even-strength chances. A .954 team save percentage is doing most of the work.
Seattle (5-2, +10, No. 28). They're getting 46% of the chances and winning on 10.2% shooting and .943 goaltending. That combination rarely lasts.
Ottawa (5-2, No. 21). They were at 55.7% of even-strength chances last season and are at 45.9% now. The model only sees the second number. Seven games isn't enough to say the team changed, so I'd rank them higher than this.
Edmonton at No. 25
This is the ranking I trust least. Edmonton is 5-0-1 at +11 with 54% of even-strength chances. Part of their low rating comes from how one half of the model reads their high-danger chances and win rate, which we're looking into. On the underlying numbers I'd put them in the top ten.
Philadelphia
The Flyers are 0-5-3 with 33.9% of even-strength xG, the worst in the league by four points. That's 17 points below last season, the largest drop of any team. They're shooting 3.8%, so the record is worse than the play. The play is still bad.
Why early standings mislead more than they used to
Across 2021–26, the spread in true team quality narrowed every season. We measured it as the standard deviation of win percentage after removing what chance alone would produce. It fell from .117 to .062. Last season also had the most overtime games (24.8%) and one-goal games (43%) in that span. When teams are this close, a hot goaltender or a cold shooting week moves the standings more than it did five years ago. Shot share is slower to move, and it has been a better guide this early in the season.
Data through October 14. Ratings from the power rankings on nhlforecasts.com; EV xG from our neutral xG model; records count OT/SO losses separately.