A 2025 Casual Fan's Guide to Advanced Stats and Hockey Analytics
Into the mathematical breach once more: this time with an aside on the Jason Robertson conundrum.
If I’m obnoxious/pretentious/obscene about the importance of analytics and stats, it’s because I always get the sense that this stuff is still kept at arm’s length from fans. The cheeseheads babbling about “will over skill” don’t want to do more homework, and so we still have fans who either don’t care or don’t like hearing about expected goals. Even those who have come to accept them haven’t really embraced the information. Instead they’ve agreed to shake hands with a stranger.
That’s fine, by the way. You don’t have to care about analytics. You don’t have to use “per 60” in a sentence. You’re a cooler person for never having uttered those words, believe me. But there’s no reason for this information to be in the margins. There’s no reason for us to speak different languages when it comes to a sport that only has one: the language of the puck.
There’s an evergreen note I leave at the top of each explainer card, and I want to repeat it here because it’ll come in handy given some of the new work being done — especially for those who have a limited view of these stats, whether intentionally or not.
“Analytics” are not mere numbers. They’re models: ways of bridging practical (what happens on ice) and theoretical knowledge (what will or can happen based on prior events). They help define the conditions of performance over a period of time rather than merely reference a series of outcomes the way plus-minus does. The climate versus the weather. These are not meant to be a right or wrong way to look at a hockey game; they’re meant to be useful in understanding the conditions of hockey itself.
Adding to this, perhaps the most important part of analytics is being able to identify when those conditions change. For example, if we want to know when the slapshot died, we can turn to analytics, because it turns out — we have an exact date. But what about other on-ice conditions that have changed over the years?
Thankfully, hockey fans have access to people who watch a lot of hockey, want to learn more, are willing to put in the work, and thus have some answers to these questions. Let’s begin with one recent hockey change: the fact that players are missing the net more than ever.
Micah Blake McCurdy of HockeyViz put together his latest model at the end of May. There’s a lot in there. While there is plenty of mathspeak, there’s also a lot of insight: the way shot outcomes happen or don’t happen, the types of shots that are more likely to miss the net or not, get frozen by the goalie, or become blocked and the gamestates that associated with shot outcomes. Instead of sifting through all that — this is a casual fan’s guide, after all — the biggest change in his new model is trying to estimate the value of rush versus cycle shots, and the variability within them. One of the league-wide trends we’re now seeing is that not only are shots missing the net more than ever, but rush and cycle chances are a lot less likely to score (blue line)
This is a little too shorthandy as far as serious analysis goes, but I do think something like this illustrates the problem that a team like the Dallas Stars had. Dallas was a very good team. But even very good teams can have a few missing ingredients. In their case, they were very good on the rush; for most of the season one of the best. But if the league is trending in the direction of more overt possession and passing plays instead of counterattacks and speed — than doesn’t it make sense that the Stars would struggle? If DeBoer was too committed to playing a certain way, then doesn’t it follow that perhaps he failed to adapt?
Again, this is less about finding answers, and more about being able to ask better questions. Instead of listening to some hack drone on and on about whether or not Dallas Star players had big enough cojones, we can investigate hockey reasons for why Dallas lost critical hockey games.
This idea of physicality is not lost on the chart huggers, though. And that’s where Louis Boulet’s work — thanks to the singular Corey Sznajder — stands alone. What does it mean to call someone physical? Josh Anderson is physical. Does that make him a good player? Even Canadiens fans would argue otherwise. Being physical is not an inherently good trait unless it’s leveraged for something positive. If that were the case, Ryan Reaves would be a regular.
Using data from Sznajder and MoneyPuck, Boulet has built what he calls the Battle Grid. This grid has an I, Robot three laws vibe, where we want to define physicality in a way that captures something useful; something that is actively leveraged instead of something that merely exists. After all, you don’t need to be able to fight and enact revenge to be tough. Peter Forsberg didn’t fight, but he was one of the most hard-nosed players the game has ever seen.
In Boulet’s case, the data is looking to turn physicality into something measurable; something more nuanced. This ‘Battle Grid’ seeks quantifiable data related to events involving contact between players, events that occur in hot zones (highly contested areas of the ice), and plays that improve team territory.
There are six categories that fall under this rubric:
on-puck contact (think about a hit on the puck carrier, or blocking a shot)
transition against the forecheck (a successful puck recovery under pressure)
checking engagement (puck recoveries and disrupted zone exits, for example)
offense created off the forecheck
playmaking down low (since passes behind the net are more successful than any other pass leading to a goal)
net-front presence, which includes deflection, tip, and rebound volume
For defenders, these values are inverted. I thought it’d be interesting to look at a real-world example: something that would prompt debate. So why not the most contentious Dallas Star of the season, Jason Robertson? He would be the perfect example on any day. Here’s a forward that plays physical without being a power forward; a forward who seems to be capable on both sides of the puck, and yet I don’t believe fans (or nerds for that matter) are truly convinced.
How does he fare on the battle grid? This season: not well.
While this visualization does reveal how Robertson’s defensive sausage is potentially made, it reveals even more some of where the fan criticism was warranted: essentially, Robertson was responsible in the defensive zone, but that responsibility didn’t extend beyond the defensive zone. He was poor at winning territory in the opponent’s zone, and just downright awful at applying pressure.
Now, does this make Robertson a poor player? Of course not. Again, this is a very specific dataset about player checking, and their net rating for and against it. This is an important distinction because someone like Reaves (sorry to keep using you as example, Mr. Reaves) actually rated higher than Robertson as a forechecker this season.
This takes us back to the importance of analytics: the power here is in precisely that they don’t have convenient or easy answers. They’re about adding language, and asking questions. (Apologies for sandbagging this next card) After all, if we go back to last season…
…Robertson was excellent in these categories.
Not only that, but the same pattern was true of Roope Hintz, who tends to get more credit for his defense.
Again: better language, better questions.
What I find important about this information is just how counterintuitive these things can be. Lian Bichsel was a revelation this year. His physicality and talent on the blueline felt like a boon — like an inclusion of something the team hasn’t had in decades. And yet would you be surprised to learn that his defense against rebounds, tips, deflections, and chances in the crease was quite poor? There are a number of ways to answer that question, and yet none of them would be useful. After all, we’re talking about a rookie. If anything, we should be asking questions. Is raw physicality the best defense against chances in the crease? Is something about Bichsel’s positioning making him more susceptible to forwards generating rebounds?1
Of course, we also know that Dallas’ team defense was really poor under DeBoer this season. A lot of Dallas players rated poorly in these categories. Is there a systems issue at work here? If so, why did it show up in one season under the same coach, but not the other?
I’m not sure we have the answer to these questions, but we do have a tool to help answer them, and there’s no reason to be afraid of them.
Explainer card index
For those who are new, you may have noticed specific captions for any time I post a picture or chart. In those captions are links to notes where I wrote about what they are, what they measure, and how they’re different — all in terms anyone can understand.
While the explanations themselves aren’t perfect, it is, I would like to think, a start. And don’t worry, normies — the action on ice is just around the corner. Even if it doesn’t feel like it.
Synthetic Goals (sG) by Micah Blake McCurdy
Setting by Micah Blake McCurdy
Goals and Wins Above Replacement (GAR and WAR, respectively) by Josh and Luke Younggren
Regularized Adjusted Plus Minus (RAPM) by Josh and Luke Younggren
Delta Fenwick Save Percentage (dFSV%) by Josh and Luke Younggren
All Three Zones tracking data by Corey Sznajder
Mitchell Brown and Lassi Alanen’s prospect tracking data
JFresh’s Hockey Cards
Deserve to Win O’ Meter by Peter Tanner (MoneyPuck)
Net Rating by Dom Luszczyszyn
Cody Ceci’s Battle Grid card might explain a lot, not to mention, why coaches fall for his illusion of defense. These grid cards consist of five seasons’ worth. Every season, Ceci manages to rate as defective in transition contests and proactive contact. His size and speed (which he has) make for a hollow counterargument, but one that coaches seem to think he can play his way out of. He’s defensive quicksand in otherwise.
One of the best analogies I've heard recently for analytics/models is that they're like a map. No one expects a map to include every detail of the terrain, but no hiker would ever refuse to use a map just because it doesn't contain every stone and branch that falls across a trail. Conversely, just because someone has explored a small area in depth and has detailed knowledge of the terrain in that area doesn't mean they know other places or can situate their own region in a broader context (a frog in a well cannot imagine the sea, etc).
Love this stuff. I don't always get it, initially, but the explainer cards are useful, everytime. Keep laying this stuff out there, and I'll keep picking it up.