Written By Ben Remington (ZoneCoverage.com)
In another installment interviewing unique Minnesota Wild fans, I sat down with Luke Solberg, better known as @EvolvingWild on twitter, one of the foremost names in advanced stats when it comes to Minnesota’s hockey team. I joined Luke in his laboratory, a sterile room with many servers and beeping lights, and I had to leave my chocolate milk outside.
OTT @ MIN – 01/22/18
Final Score: OTT 1, MIN 3
Cumulative Corsi/xGF (EV) – score & venue adj.
Box Score tables – EV pic.twitter.com/BCJIlhiOWR
— EvolvingWild (@EvolvingWild) January 23, 2018
Ben Remington – Thanks for joining me. I’ll just jump right into things here. You’re the famous @EvolvingWild, purveyor of superb Minnesota Wild statistics and visuals. From the beginning, what got you there? Have you been a Wild fan from the beginning?
Luke Solberg – First of all, thanks man! I just like making charts and working with numbers. And the Wild of course.
I’ve been a Wild fan since I’ve been a hockey fan, which started around eight years ago. I was a Twins and Vikings fan from a very early age and became kind of obsessed with the Twins/baseball in general in high school. When I graduated from college, however, I had a lot of free time on my hands, the Twins were terrible and I had always been interested in the Wild/hockey from a distance. My dad watched them often and played hockey — he still plays hockey for that matter — so I figured why not?
After five games or so, I was hooked.
It was pretty natural for me to immediately go to the more “advanced side” of the game when I first started watching given my love for baseball and sabermetrics. I learned what Corsi/Fenwick/etc. were a few months after I started watching the Wild, and I’ve been obsessed with the “numbers” side of the game ever since.
BR – So since you’re relatively new to hockey, but have always been a numbers guy, do you think that helped you learn the game? Do you feel like you’ve got a pretty good grasp of hockey on the ice as well as in the numbers?
LS – I wouldn’t say it necessarily helped me learn the game. Growing up in Minnesota, there are a ton of people who “know” the game, and those people love talking about hockey. I learned the game from watching with more knowledgeable fans/viewers — asking a lot of questions, trying to get the rules down, the flow, high-pressure situations, the best players, all that stuff. I was also a big Reddit/Twitter lurker for a long time, which also helped in a big way.
For me, the numbers just made a lot of sense. Back then, shot attempt-based metrics were the only thing available for the most part. They kind of still are. The idea of CF%, for instance, just seemed like a logical and normal thing. I thought that outshooting your opponent consistently should absolutely play into how teams/players were viewed and valued. If a team gets outshot or out-chanced etc. on a regular basis, it made perfect sense — to me — that we’d maybe view them as a weaker team or vice versa.
At this point, I’m very confident in my eyes, but we’re only human. There’s a lot more going on than we can possibly process in a hockey game, at least in my opinion, and I’m constantly training my eyes based on what I see in the data and the work I do.
BR – I think you touched on an important piece there, about your eyes being human. Confirmation bias is a big reason why advanced stats are needed, I think. Do you think statistics can change peoples’ biases?
LS – Only if people are willing to believe that what they see might be biased, or rather, the way they see it. I know that stats/numbers aren’t the only things we can look at; there are a ton of factors that we’ll never be able to track or measure. However, I’m often surprised by what I thought I saw and what something based on the data says. For me, this is the way I try and balance the natural confirmation bias we have as fans. So yeah… I trust the statistical approach to hockey a lot, but there are definitely situations where I’m skeptical, or I’m willing to give a certain player time, for instance. It’s definitely a push-pull kind of thing.
That being said, I 100 percent think statistics can — and should — change peoples’ biases, but there are plenty of people who would rather focus purely on watching the game and judging/evaluating that way. I understand that mindset, I just don’t really agree with it. If we have information that might offer a different viewpoint or different way of evaluating something, I think it’s a bit insane to ignore that.
BR – I would tend to agree with you on that. I think that statistics should back up what I see. I played organized hockey for years, and fascinated myself with the techniques. But I can’t imagine watching a game, and wanting to know more, without looking at any numbers.
There are still plenty of critics of advanced stats in the hockey world, though, and some of their arguments do have merit. What are some criticisms of advanced stats that you hear that you think are valid, and as a number-oriented guy, how do you overcome them either as a fan or when analyzing the numbers?
LS – Well, I think there are two groups who tend to criticize analytics: those who don’t like using them– the eye-test crowd — and those who really like using them (people like me). I’ll just stick to the first group for now. The biggest criticism I hear from that group resembles something like “this guy does more than the numbers could possibly show” or “you can’t measure everything” etc. I think there’s obviously something to that. It’s easy to get caught up in the numbers and forget there are real-world things that influence a player’s performance, or a team’s ability to win, etc.
But, at the same time, most of those things — along with the idea of leadership, grit, intangibles — are immeasurable or unknowable. They should definitely be taken into account, but I don’t think people like you and I or your standard fan on Twitter should be the ones to do that. We’re not in the organization, we don’t know these players, etc. In general, I think it’s always worse to assume — because you heard a story about a player somewhere in the paper or online — that you now know the situation and can make an informed decision about said situation.
So I generally just stick to what we know.
The other part of that, the “you can’t measure everything” is a very legitimate criticism because it’s true. You can’t measure everything. However, we have a TON of data to work with, and I generally believe that a good (objective) model, for instance, will do a better job of evaluating a player than any one person “watching the game” ever could. Of course, that assumes that we have enough data for any given player. There’s still incredible value in scouting — I’ll never argue that point. Personally, I more or less ignore the numbers for a good chunk of a rookie’s first season, for instance. But overall, a solid blend of scouting and analytics will probably yield the best results. It should never be one or the other.
BR – Again, I’m with you on that. There’s always got to be balance in player evaluation. On the other side of the coin, there are plenty of criticisms that are, let’s just say, not very legit. What are your favorites that you’ve seen in that genre? Also, feel free to expound on how ridiculous they tend to be.
LS – Oh man, let’s just say I have several. I’ll try and keep it easy for now though. There are so many “Corsi is awful” or “Corsi doesn’t really tell you what Kris Russell does for this team” type things that get thrown around in earnest, and they still really irk me. Mainly because those people turn right around and use traditional plus-minus — or total blocked shots, etc. — as a player evaluation tool.
Plus-minus is garbage. Even something like on-ice 5-on-5 goal differential is suspect at best. Everyone relies on statistics to some degree, and if you’re going to criticize “advanced stats” but you’re somehow okay using decades-old box score metrics that have been shown time and time again to be more flawed — and worse — than the newer metrics… I just don’t understand. Goals are obviously how teams win games, but there’s a lot of things that go into how a goal is scored that a skater has no control over.
I could rant about this forever, but I don’t understand how people are okay with using one statistic and not another, especially when those people criticize the other so vehemently. And often, there’s no actual argument to back up their disgust, it’s mostly just ignorance that fuels the criticism. To be honest, I have a lot more respect for guys who dislike the advanced stuff, but also don’t cite anything else. Someone like Lou Nanne comes to mind. They literally just watch the game, that’s it. I don’t know, I could go on forever here.
One last thing I’ll say, if you’re going to criticize something, at least put in the effort to understand why others feel it’s useful — or understand it to begin with. If you haven’t done your research, I have no time for your criticism.
BR – Well, that answer went about how I expected, so thank you for that. I think the criticisms surrounding advanced stats not working, or not being able to build winners is silly as well. We’ve seen the Wild delve into the world of advanced stats, hiring the War-On-Ice folks, but it hasn’t brought them more success than years previous.
At the same time, every move Chuck Fletcher made this summer seemed to be the total opposite of what advanced stats would tell you to do. If Wild fans want to shake their fists at advanced stats, no one is going to stop them, but it also seems that the team isn’t using them as often as they should be, either.
Do have any thoughts on how it seems the Wild are using analytics?
LS – I wish I had actual information about how the team is using their analytics department. A.C. Thomas is one of the most well-respected statisticians in the NHL, and their department, based on everything I’ve heard, is definitely a part of the overall plan in the organization. But you’re right, Fletcher’s moves this past offseason made no sense from an “analytics” point of view. I guess I could see the Foligno part of the deal a little bit, but even then you’d expect them to stick to a <$2 million AAV for 2 years max kind of thing.
I think Fletcher has relied on them before — there are a couple trades/transactions that just scream it. The Eric Staal signing is a big one. He was VASTLY underrated by the market when he was a free agent, and look where he’s at now. That’s probably one of the better contracts in the league right now.
The Dubnyk trade was another great move. He sells it like it was pure luck, but there were things there that indicated Dubnyk might turn into what he’s been for the Wild. Also, the Spurgeon, Nino and Granlund signings are in agreement with what the numbers show. Those are all very good contracts. The Wild have also drafted really well in my opinion — with what Fletcher left them — as Eriksson Ek, Kunin, Greenway all have loads of potential.
So there’s probably something there too. So yeah, he’s definitely using the department. I mean why wouldn’t you when you have one of the smartest guys in the field working for you?
* The Dubnyk trade may have pre-dated the new analytics department, but the point probably still stands
With all that said, this summer shows that Fletcher will sometimes allow himself to turn a blind eye to the numbers ([cough] Jarret Stoll). I mean he signed Kyle Quincey… a left-shot “fourth pair” D to play the right side. And the Buffalo trade is baffling. To be honest, I actually liked trading Scandella. I was a big proponent of moving him in expansion or the offseason, but also moving Pominville for Tyler Ennis & Foligno?
It just didn’t — and still doesn’t — make any sense. Why not move those players for whatever picks you can get to free up cap space?.
The team’s loss to St. Louis clearly influenced their moves this past year, but I also think Fletcher was planning on getting Kaprizov, and it also appears Parise’s injury came out of nowhere. If I was forced to speculate, I would say Fletcher relies on his coach’s opinion above almost everything else. If his coach wants a fourth-line center who can win face-offs, Fletcher gets Stoll. If his coach thinks his team isn’t tough enough, Fletcher gets Foligno and Quincey.
But that’s pure speculation.
BR – That’s a good point, that a coach’s input will likely trump that of an analytics department, and perhaps rightfully so, since his keister is on the line more than theirs.
I can’t thank you enough for taking the time to chat with me, and giving me such well thought out and detailed answers. One final question, where do you see hockey analytics going right now? It seems to be following a similar trajectory to sabermetrics in baseball, obviously just many years behind. What would you like to do and see with hockey advanced stats in the future? Can I please get goal net heat mapping before I die?
LS – Thanks for having me. These were really great questions. It’s been my pleasure! Hockey analytics is in a really interesting place right now. Hockey is similar to baseball in that a lot of people are still hell-bent on convincing the mainstream, so to speak, that this stuff is important — more or less.
At the same time, the technology we have available — not to mention the computing power compared to say 15 years ago — is nothing like what baseball sabermetrics was built on. The analytical side of baseball in the public sphere, for all the fame and coverage, is still very much pen and paper. While it can get VERY complicated, something like WAR has a basis that is made up of relatively simple math. Hockey, on the other hand, has moved into a realm of statistical modeling and data science that’s really not present in the public baseball stats you often see. It’s similar to the advanced metrics seen in basketball — APM, RAPM, etc.
In hockey — with the combination of the technological advances and the large number of things we haven’t explored — we’re seeing more and more new work being done using machine learning. Hell, just go read Emmanuel Perry’s write-up on his game prediction model “Salad” or Micah Blake McCurdy’s “Edgar” model — they’re nuts. The techniques used are so much different than what sabermetrics did when it was in a similar place. FIP, wOBA, wRAA, wRC+, for instance, are ‘pillars’ of baseball analytics, and they’re relatively simple metrics in comparison.
That being said, an increase in complexity does not necessarily equal an increase in quality. Some may argue that as long as an increase in complexity means a “better” model by a given evaluation metric, the increase in complexity will always be better. To me, it’s a balance. The simplicity in baseball is quite beautiful.
Granted, baseball is a much easier sport to capture with numbers.
I’m kind of rambling here, but my main point is: there’s still so much to be done with the data we have and the technology that’s at our disposal. I see this “all the low hanging fruit has already been picked” opinion from time to time, and it’s become apparent to me — after working with the NHL’s play-by-play data — that this couldn’t be further from the truth. There are so many things that haven’t been explored.
I mean, we’ve had, what, three WAR models in the existence in the NHL? And only one is actually available right now — Perry’s on corsica.com? Is there anything close to a site like Fangraphs or baseball prospectus? I think passing and zone-entry tracking — like Corey Sznajder’s and the passing project’s work — is really important as well. Some great people are doing special team tracking/analysis, there’s a pretty big hole in proper contract projections/salary analysis, and who the hell knows what’s going on with goalies?
The list goes on…
I’m not sure where hockey analytics will be in say five or 10 years. I think we may see the player tracking data that’s been rumored for a while, but I’m not sure we’re ready for that.
I mean I’d say we’ve done like maybe 40-50 percent of what we can with what’s currently available. I’d like to see the NHL go all in on the passing project and give us data on at least each pass leading to a shot in every NHL game — with coordinates and time data — before they do anything with actual player tracking. Maybe even pick up Sznajder’s zone-entry work. Those are my wishes.
Regardless, the future is bright.
BR – Thanks so much for doing this. Hopefully, you’ve given fans a bit more of an introspective look into hockey analytics. Also, thank you for all the work you do crunching the numbers and sharing them for the Wild. Providing content like that can be a bit of a thankless job sometimes, but you’ve rightfully earned a big following and I’m sure I speak for the rest of your Twitter followers when I say that I very much appreciate it.