The Seattle Kraken’s signature win on Monday against the Carolina Hurricanes marked Game 60 of the season, which means it’s time for another 10 for 10. If you’re new here, this series pulls together 10 data points from the last 10 games to help show where the Kraken are trending, what’s working, and what still needs attention as the playoff race tightens.
Let’s jump in.
Data Point 1: How they fared
The Kraken went 7-3-0 over their last 10 games, beginning on Jan. 25 against the New Jersey Devils. When you compare that stretch to the other Pacific Division contenders and the wild-card hopefuls, you can see how they’ve positioned themselves firmly in the playoff race.
Data Point 2: Change over the season in the Pacific Division
One fascinating trend this year is the shifting landscape of the Pacific Division. The top five Pacific teams from last season have all seen significant dips in points percentage, while the bottom three have taken major steps forward and are now in playoff contention. No other division has seen a swing this dramatic.
Data Point 3: Holding on to leads
This chart is a bit of an eye-sore, but I stumbled across it over the weekend and couldn’t look away. The Kraken have won 100 percent of the games in which they held a two-goal lead at any point. They’re one of just three teams with a perfect record (23-0-0) in that situation.
If you look one column to the left, their 69.0 percent win rate when holding a one-goal lead is right around league average.
Data Point 4: Goals against (minus empty-net goals)
We knew in the offseason that new head coach Lane Lambert would bring a more defensive mindset and structure to the 2025–26 Kraken. That was on full display early in the season as Seattle squeaked out low-scoring wins by minimizing goals against. Things slipped a bit in Games 21–30 and 41–50, but the last 10 games have been outstanding in terms of limiting goals allowed.
Data Point 5: Goals for (minus empty-net goals)
Goal scoring remains an area of opportunity for the Kraken. They’ve struggled for much of the season, and it’s been a recurring theme in national media skepticism about their playoff chances.
Their 2.9 goals for (minus empty-net goals) over the last 10 games is still in the bottom half of the league, but it’s encouraging to see improvement.
Data Point 6: Goals by player (excluding empty-net goals)
Digging deeper into the scoring, Shane Wright and Matty Beniers led the way over the last 10 games with four goals each, excluding empty-netters. Jordan Eberle scored five goals in that span, but two came into an empty net.
Data Point 7: Home is where the wins are
If you’ve been attending games this season, you’ve probably felt a little extra pep in your step walking out of Climate Pledge Arena.
It gets better: in calendar year 2026, the Kraken are 8-2-1 at home.
Data Point 8: Winning in the division
The Kraken are also showing up against their Pacific Division opponents, posting the best points percentage within the division.
Data Point 9: Expected to lose
It’s no secret that much of the national media is betting against the Kraken making the playoffs. With so many analysts based in the Eastern time zone, I’m guessing they’re not staying up late for Kraken games, and if they are, they might be more inclined to watch the flashier West Coast teams like the Sharks or Ducks. I’ve got no problem with that, but it does mean they lean heavily on publicly available advanced analytics.
One category they often rely on is “expected goals” models. For those unfamiliar, expected goals (xG) estimates the likelihood that a given shot becomes a goal based on factors like shot location, shot type, pre-shot movement, game state, and more. They’re useful directional indicators of how well a team played, regardless of the actual result.
In a single game, the sum of expected goals for minus expected goals against is often used to determine who “should have” won. Over the season, these accumulate into an expected goal differential.
The Kraken rank 29th in expected goal differential but 17th in actual goal differential (excluding empty-net goals). Knowing how these stats are used, it’s easier to understand the national pessimism.
Public models are great tools, I use them all the time, but they’re incomplete and can sometimes be misleading. I’m not declaring this a playoff team with certainty, but I’ve long believed the Kraken do something different that public models don’t fully capture.
Data Point 10: Signature win and shot suppression
This 10-game segment closed with a signature win in Game 60, when the Kraken beat one of the NHL’s best teams, the Carolina Hurricanes. Some fans expressed concern about the shot count, but I felt comfortable with how Seattle was playing. After the first period, I thought they had a formula to win. Sure enough, they scored twice in the second (and had a third goal negated by an offside challenge) and held on for a 2–1 victory.
The reason I believed they had a shot was the way they limited Carolina’s quality opportunities. To illustrate, look at the Hurricanes’ shot attempts over their last four games.
It’s clear Seattle did the best job of limiting high-danger attempts in front of the net. Don’t get me wrong — it wasn’t easy, and Joey Daccord still had to play a terrific game, but it was a full-team effort to suppress Carolina’s chances.
Taken together, these 10 games showed a team sharpening its defensive identity, finding just enough offense, and banking points at the right time. There’s still plenty to prove, but the Kraken continue to hang around and make things interesting.
If you remove empty netters from the actual goal differential, should you remove empty net attempts from the expected goal differential? Would that produce any notable changes to the rankings? I know Natural Stat Trick has with/against empty net filters.
Coincidentally I also looked at Dom’s dubious playoff “projections” today.
Having worked at a Wall Street firm for many years I probably have a higher tolerance for superflous, model-based quackery than most. But, geez give me a break. He has the Kraken finishing with 85 points, whereas they currently stand at 67 points with 22 games left. To “achieve” that means the Kraken would capture ~ just 41% of their remaining points. I can be as pessismistic as the next guy, but I’m taking the over. The cherry on top is DL’s projected goal differential – he has the Kraken finishing at -27 (excluding empty netters I believe they are currently at 0).
DL needs to take his model somewhere and “recalibrate” it. Preferably in the nearest trash compacter….
Dom’s model is all xG, all offense and anyone who knows sport knows that defense matters… but whatever.
Incorrect. The Athletic model rates defense, as well. And they track how it performs relative to other prediction models and it has consistently been among the best. A lot of “shooting the messenger” going on here…
By “all” I mean heavily. The situational components are also extremely muted. Furthermore, Dom’s model does fine on the aggregate but it struggles with certain specifics. Last season’s Kraken are a good example. His projection was only off by 11 points… not bad. That was actually the best showing he’s had on the Kraken ever. The problem is, his model projected a middle of the league defense and a bottom of the league offense. Those were exactly backwards. The model was wrong even if the conclusion was correct.
To draw from the famous George Box quote, I agree some models are useful, but all are wrong. Some are more useful that others, and some are more wrong than others, but it also depends on how you use them.
I think it’s safe to say the Kraken will do much better than his forecast 29th in the league finish. Of course part of that is his model – and a lot of other models – missing so badly on the entire league.
John, the charts are really nice and much appreciated. Some interesting tidbits here.
But the talk about “east coast bias” and “they don’t watch the games” is beneath you. Come on, man. I just wrote this on an old thread, but I’ll put it again here: I would say there are only 2 teams in the entire league without at least ONE player better than ANYONE on the Kraken: Vancouver and Calgary — two teams that are actively tanking this year. That is assuming one is willing to write off Huberdeau as washed up… which I am. Please talk me into any other teams. I know elite players don’t grow on trees, but after 5 years, that is abysmal. And this is a “strong link” league. THAT is the main reason everyone is so skeptical of this team. People pay the most attention to star players, and the Kraken simply don’t have any.
As for the idea that the Kraken do something special that the “public” models don’t capture (and models like the Athletic’s are NOT public, only the results are) — what do you think that special sauce is? Isn’t it simply goaltending? Expected goals don’t pretend to capture goaltending, and the Kraken’s goaltending has been really good, plus mix in a little favorable puck luck. It’s really pretty simple. That is what any of the stat-heads will tell you. And it seems right to me as someone who IS on Pacific Time and watches the games. The Kraken have hardly ever held the run of play this season but have been bailed out many nights by really good goaltending.
Actually, the Athletic’s model is pretty public. Dom’s model (the Athletic’s model) is pretty well spelled out. His “Net Rating” is detailed in the article in the link below if you have paywall access. It seems to me the difficulty of evaluating defensive is addressed by a heavy reliance on expected goals and that in itself, to me, creates something of a special circumstance for a team that plays low-scoring, situational hockey.
https://www.nytimes.com/athletic/4396412/2023/04/12/nhl-advanced-stats-offensive-defensive-rating/
Why would that be? A team that plays good defense, suppressing shots and keeping them to the outside, would be *allowing* fewer “expected goals,” which would be reflected in the model.
I think, and this is speculating, the Kraken keep shots to the outside but that results in a high volume of low quality shots. They especially seem to do this when they’re leading and I think that’s a significant factor in why they win more often when they are outshot by their opponents. This volume of low quality shots results in a cumulative negative expected goals share even though they actually win those games. This, to me, is where I wonder how accurate the models actually are at capturing not just the location of the shot and the pre-shot movement but also the game state. They’re supposed to do this but I’m highly skeptical they can actually capture this level of complexity, especially in an effort to quantify defense.
I think the Kraken are undefeated in games this season after establishing a two goal lead because they play a very situational game and it seems to me that’s partially why Dom’s model has not done a very good job projecting their performance… and also much of the rest of the league.
I do wonder if parity is making this problem more apparent. I scoured over Dom’s reviews from last season a while ago and found he was off by about 25% on 25% of his projections. That seemed like a lot to me, maybe it’s not. I expect this season will be worse. Anyway, he was much more accurate on the obviously good teams. One would think if the model is useful, it will need to do more than tell you the good teams are good, it will need to tell us which mediocre teams are actually good of bad. In the case of Seattle, it hasn’t done that.