2026 updates to the NHL team salary cap rankings

The end of the offseason is nearing, which means it’s time for everyone’s favorite (or least favorite) series at Daily Faceoff: the salary cap rankings, grading every team by their financial situations.

The rankings are where I put together a comprehensive evaluation model to (as objectively as possible) rank every NHL player signed to a contract and then compare them to their salary to determine how much value they bring (or take away) from their team relative to the market. I then use those player contract ratings along with the number of no-trade/move clauses a team has, their dead cap space, the quality of players locked up to contracts four years or longer, and their cap space relative to the quality of their team to determine how well every team manages their salary cap picture.

Normally, I use this article to break down the entire system I use to determine these rankings, but that was also when I used tiers (which changed ranges every year) to evaluate players, and then last year when I completely overhauled the system to incorporate contract ratings. However, since the system has largely stayed the same, at this point it’s easier to just redirect you to the previous edition.

But that doesn’t mean the system is the exact same again this year. While nothing was done to significantly alter the system, I figured I would instead use this primer to mention the updates I made. For the people who follow this series year after year and know the process, this is a great way to find out what’s new without having to wade through the entire summary again. But for those new to the series or those in need of a refresher who will instead look at the previous edition, don’t worry, I will also be updating that article to include the new details as well, so you’ll learn about them as we go along.

With all that said, let’s look at what’s new.

The changes to GAR (and the introduction of xGAR)

Previously, I would separate the goals above replacement stat into offensive, defensive and penalty sub-stats that Evolving Hockey provides for its model. However, with the addition of expected goals above replacement (xGAR) to the model this year, instead of having six different sub-stats to keep track of, I only incorporated their cumulative stats for the sake of simplicity.

xGAR is similar to GAR, except instead of using the stats to evaluate a player’s impact on the result of the game (a.k.a. the goals), it evaluates a player’s impact on the process of the game (a.k.a. the things leading up to the goals). For some reason, I never used this stat, but after learning more about it, it’s become a useful tool in lots of my work, including my player evaluation model.

Additionally, GAR (and xGAR) will not be tracked as a rate stat like I had in the past. Instead, I will use the total stat. While the rate stat would further reward players who played well in limited minutes, the total stat rewards the good players who’ve played more (and punishes the bad players who’ve played more), and means we won’t have as many random players with three games played rated as well as some of the game’s elite players.

Which players qualify for the cap space to skill differential category

When I calculate the team skill rating for the cap space to skill differential category, instead of incorporating every player under contract, I only use the 20 players (12 forwards, six defensemen, two goaltenders) most likely to play a majority of the season on the NHL roster. In the past, I would just take the players with the highest average ice time in their respective positions on each team, and use those players.

However, I’ve decided to adjust how I determine which players to include. Instead of going off average ice time, I now use the projected lineups on Daily Faceoff to determine the players used for this category. While this could simply be boiled down to using in-house resources where I can, it’s also because our projected lineups are often a combination of previously established lines, stylistic matches for the new additions, and information from team sources about what they expect the lineup to look like.

This change will give me a more accurate lineup to go off of and in turn create a more accurate projection for their team’s skill level for the season. However, this also means players with no NHL experience need to be incorporated into the fold, which means I need to solve another problem to fully incorporate this change.

Evaluating players with no NHL experience

In the past, my qualifications for which players were evaluated in my model came down to two factors: they had to be signed to NHL contracts, and they had to have at least one regular season game of experience. This allowed me to only incorporate the players with the stats used for the model, and allowed me to maintain objectivity for it.

However, there was one issue with it. The teams who took NHL-ready talents in the recent draft, or were projected to have NHL-caliber prospects make their rosters out of camp were punished, especially in the Cap Space to Skill category. They were expected to get potentially high-end talents added to their teams, players who could significantly improve their teams, but in terms of their skill ranking in my model, they were ignored.

To correct this, I’ve decided to finally incorporate rookies into my model but only for rare circumstances. With the new method to determine which players make my “projected lineup” for the skill rating, any rookie without NHL experience listed on Daily Faceoff’s projected lines will now be added to the model. For this year, the only players who met this requirement were the Chicago Blackhawks‘ Roman Kantserov, the San Jose Sharks‘ Ivar Stenberg, and the Toronto Maple Leafs‘ Gavin McKenna.

In order to incorporate these players as objectively as possible without the information every other player has, I instead went off of the scouting reports of these players and looked at their NHL player comparables. I then used these comparable players, collected the data from their respective rookie seasons and combined that data together to create the projection for how these rookies may play this season.

This gives the model something to go off of in terms of their impact for the team skill rating, as well as how much value they’ll provide relative to their entry-level contracts. And of course, when I come back to do this next year and they have some NHL experience under their belts, they’ll be treated like every other player in this model.

Reminders of the contract rating system

This one is less an update, and more of just an annual refresher of the contract rating system to remind people about the intricacies of some of the results, and to avoid as many overreactions to the results as I can.

Since this model is looking at the past three seasons of data, players who may have been just okay in 2022-23 and 2023-24 but broke out in 2024-25 will be hampered by the model, even if that recent breakout is the new established level for that player and they were signed to a contract as such. Since I don’t have the time to incorporate projections for players down the road, both in terms of younger players getting better and older players getting worse, it is one of the few flaws in this process.

For example, Juraj Slafkovsky’s 212th ranking doesn’t line up with his current level of play, but that’s because he’s only been at this level for about a season and a half, so half of his data in this model is still incorporating the Slafkovsky who didn’t look close to a first-overall pick. That combined with his still-existing defensive issues punishes him more than it should, even if he’s still risen up significantly from his 341st ranking last year.

As an opposite example, the model finally recognizes Elias Pettersson as a weaker talent with a 148th ranking, because the sample of him playing poorly now makes up two-and-a-quarter seasons of the data used in the model. Compare that to last season when it was only one-and-a-quarter, and he was rated seventh, it’s a significant change, but the patience is needed to get an appropriate sample size of it.

This also means that players signed long-term to deals which are projected to age well aren’t reflected as such in the model, since it is more of a descriptive model than a projection model. Take Cole Caufield, for example. Like many of the Montreal Canadiens‘ core players, his $7.85-million cap hit with five years left is considered to be excellent value by public perception, especially in the later years. However, right now it has a rating of -6. But that doesn’t mean it will always be a market-value deal. In fact, it being at market value now means that as the cap continues to rise, the value for the deal will increase as more contracts become more expensive than it. So long as Caufield doesn’t get worse, it will be reflected in the coming years.

_____

SPONSORED BY bet365

Recently by Scott Maxwell

• Five NHL players who could thrive with their new teams in 2026-27
  The five best bargain options available in NHL free agency
• What the Red Wings would look like with better lottery luck
• The top five remaining NHL goalies in free agency
• The five most mediocre NHL teams of the past 10 years
• The 10 most overplayed songs at sporting events