Game 01: Playing for 10 minutes is not enough

A deep dive into Winnipeg Jets's 4-5 loss against Dallas Stars

The Winnipeg Jets are back for the 2025–26 season and are hoping to build off an impressive Presidents’ Trophy-winning campaign that included one of the hottest starts in NHL history.

Unfortunately, this season already feels different from the last.

I will be live-posting games via Substack’s app, Twitter, and BlueSky, although Substack will get my main focus (well, as much focus as I can give it with tracking being the priority).

---

Per requests from our summer survey, we will focus on game recaps this season. This will be the first recap, so I’ll go into a bit more detail on the graphs and terms for those of you who are newer to hockey analytics.

I still plan to do a weekly post, probably on Mondays, separate from the recaps to provide more in-depth analysis on trends, along with my own updates and some random non-hockey fun.

Winnipeg Jets 4 - Dallas Stars 5

Corsi is simply all shots, whether they are goals, saves, misses, or blocks. Corsi is often used synonymously with possession, as it was noted quite early that the team with higher shot totals tended to coincide with the team Tim Barnes (under the blog pseudonym Vic Ferrari, now the Director of Hockey Analytics for the Washington Capitals) felt controlled possession for much of the game.

I personally prefer to call it “useful possession,” as — while it does correlate with actual puck possession — it specifically denotes possession opportunities that result in shots, rather than all puck possession. It has also historically been a stronger metric than actual puck possession in both descriptive success (which team won that game) and predictive success (which teams are most likely to win in the future).

We see here that the “possession” battle was fairly even for much of the game, but the Stars took over around the halfway point.

Having more opportunities than your opponent is good, but not all shots are created equal. This is where expected goals (xGoals) come in.

xGoals are all unblocked shots (so no blocked shots)* adjusted for all the shot-quality factors available publicly via the NHL’s data. This includes shot distance, angle, type of shot (wrist, slap, snap, etc.), handedness of the shooter, and also the location, time, and type of the previously tracked event.

* HockeyViz.com’s xGoal model is unique in including blocked shots. It makes an educated estimate on shot quality based on factors such as the location of the shot blocker.

There are many shot-quality factors not tracked that heavily impact an individual shot’s xG. These can include things like pre-shot movement and passing information. However, it appears that adding these factors to xGoal models only improve things marginally by about 10-15%.

Despite pre-shot movement and such being extremely important in individual shot probability, the marginal improvement is likely because shots/players/teams already excelling in the publicly available data are largely the same in the pre-shot data.

That’s fine for large sample analysis, but it does mean individual shot and even game level analysis can be a bit wonky.

Adding shot quality makes the game look more tilted toward Dallas for a greater portion of the game, in large part due to the Stars blocking many of the Jets’ early chances.

When we look at the even-strength Corsi data at the player level, we see that pretty much all of the Jets — except Mark Scheifele, Gustav Nyquist, and the Jets’ top pairing — were outshot.

That third pairing though… ouch.

Adding shot quality with xGoals doesn’t paint a much different picture.

Tanner Pearson jumps up due to small-sample fun, with Nyquist and Scheifele both falling underwater. Gabriel Vilardi drops down into Logan Stanley territory due to some rough 4v4 shifts.

But wait, where is Luke Schenn?!

Well, he didn’t fit on the default scale for the visual because he was outplayed so significantly:

This could be a problem for the Jets moving forward. With Dylan Samberg injured and Colin Miller eating press-box popcorn, the Jets have two of their three highest-impact defenders per hour of ice time absent.

It’s not just the quality of their play, but also that one of them was always on the ice when Josh Morrissey or the other wasn’t. That made the Jets difficult to play against.

However, bad luck and trying to be “difficult to play against” in a different sense have made the team a lot easier to actually play against.

Ville Heinola has done nothing to cement his position as an NHLer, but I don’t see how players like Logan Stanley and Luke Schenn do anything to earn their role… other than being large. Somehow, when you have smaller players sitting on the bench like Morrissey, Pionk, and DeMelo, you need to have someone big on the ice to make up for that size difference.

Because that makes sense. /sarcasm font

Haydn Fleury, who I view as better than Stanley and Schenn but not nearly to the degree that many in Jets fandom do, had himself a rough game with four goals against.

He probably would (and will) be a better number six than either Stanley or Schenn, but Scott Arniel needs to wake up and start dressing Miller again.

---

Anyways, that’s my quick thoughts on the game. Today’s focus was a bit more on the technical background side, but we’ll get more into the dirt and nuance in games coming.

And now to look at decision making and actions of individual players with microstats such as scoring chances, chance passes, zone exits, zone entries, forechecking, and entry denial tracked manually by moi…

Microstats Results

Each post-game analysis will include that game’s 5v5 manually tracked microstats at the end. I’ll provide seasonal updates in the weekly Monday post.

This portion will be paywalled for premium subscribers as a thank-you (from my blood, sweat, and tears — okay, a bit dramatic) for supporting the site.