ST. PAUL, Minn. — The morning of Dec. 2, Dean Evason led the Wild coaches’ usual staff meeting at the TRIA Rink.
Evason doesn’t use a separate office, as his predecessor, Bruce Boudreau, did. Instead, he has a desk in a room shared with his assistants, so they can communicate more quickly. There’s a big screen TV above Evason’s desk, and on this day, he went over clips from the previous night’s win over Edmonton.
Advertisement
One of the first topics? How the game broke down from an analytics perspective.
Enter Mat Sells, the team’s vice president of hockey strategy.
He’s the Wild’s one-man analytics staff.
“What did you see, hockey-wise?” Sells asked. “What did you see on the scoring chances?”
The staff discussed. And then the 39-year-old Sells presented the data.
“What Mat Sells does for us is he dummies the numbers down,” Evason said.
For example, they started breaking down their next opponent, Anaheim, and Sells told the staff the Ducks had the third-highest rate in the league at dumping the puck. That gave the coaches something to work with when creating a practice plan, which included retrievals and breakout drills.
When Evason presented the game plan to players, he delivered a phrase so popular that they knew it was coming.
“Via Mat Sells.”
The Wild may have the NHL’s leanest analytics department — Sells is the only full-timer — but there are very few decisions the team makes that aren’t at least partially influenced by its work. The decision to scratch rookie Calen Addison. The demotion of Marco Rossi. The buyouts of Zach Parise and Ryan Suter. Sells will be one of the first calls general manager Bill Guerin makes when debating trade-deadline targets or upcoming contracts for Matt Boldy, Filip Gustavsson and others. Sells travels with the team and is in most coaches’ meetings.
Guerin and Evason are perceived to be “old school.” And in some ways they are. But you can tell by how the organization operates that they’ve bought into analytics big-time.
“I think a lot of coaches won’t,” Evason said. “They’ll say, ‘I don’t want f—ing analytics. But why wouldn’t you? It’s amazing the tools we have now. We’d be stupid not to use it.
“They shouldn’t be the end-all. It should be a piece of how we’re coaching and what we do.”
#NHLAnalytics update 📈 pic.twitter.com/IqR9miNe9A
— Shayna (@hayyyshayyy) October 27, 2022
So who is this guy so trusted by the Wild staff?
Advertisement
“He’s a hockey guy,” Guerin said. “Who just so happens to be brilliant.”
Sells didn’t have much of a playing career, growing up in the small town of Gull Lake, Saskatchewan. He played midget hockey, then eventually senior league. How did he play? “Not well,” he joked.
But Sells was a huge sports fan — specifically a CFL fan — and would grab the sports page and dig through the box scores. He got his start in the business in 2012, volunteering to help the staff for the junior team in Penticton, B.C. The Vees, led by Troy Stecher and Steven Fogarty, set league and national records on their way to winning a national championship, their first in 26 years.
Sells’ big break came with a move to Hockey Canada, where he helped on the staff of the 2014 gold medal-winning team in Sochi. He started in development, doing on-ice skill work, from 5-year-olds up to junior. He was known for his ability to break down video, which was to his benefit when the program got grant money to hire someone who could help pre-scout for the Olympics.
It was in October 2013, and by then everyone who was very good already had a job. Sells was there. He became an assistant to video coach Andrew Brewer, spending his time cutting video of other national teams. Brewer and Sells put together an extensive library from international tournaments. When Canada’s coaches got started, they were handed a tablet with folders breaking down each team’s forecheck, power play, penalty kill.
“Everything you could think of, we had it,” Sells said.
Sells said analytics were still in the very early stages in hockey — the expected goals model wasn’t quite there. But he looked a lot at Fenwick (shot-attempt differential). He also worked on factoring quality of teammates and competition into assessments. This helped with Canada’s roster selection — not necessarily with Sidney Crosby or Jarome Iginla but the fringe guys. Sells and the video/analytics staff could show coaches and brass like Steve Yzerman how each player was used with their NHL team, their strengths, and how they’d fit. Brewer, in Sochi, would give Sells, back in Calgary, projects, and they’d be cut up by the time the Canada staff woke up the next morning.
“I was hooked,” Sells said.
Advertisement
Sells, who started law school at the University of Calgary in 2014, stayed on with Team Canada, getting a bigger role when Mike Babcock brought Brewer to the Red Wings. He was on Jon Cooper’s staff at the 2017 World Championships in Paris, recalling a bucket-list moment of having lunch with Cooper, Gerard Gallant and Dave Hakstol.
Sells joked he was trying to get Cooper to bring him to Tampa, but the late Tom Kurvers — a former Lightning exec who joined the Wild — was big on him coming to Minnesota for the 2019-20 season. Sells was hired by former GM Paul Fenton, who was replaced two weeks later by Guerin.
How does Sells sum up the journey?
“A lifetime in the game and just spending so much time with coaches and managers that I just know what they want to talk about and how to talk about it,” Sells said. “You can learn a lot in life by just listening. Soak it all in.”
It was late July 2021, and Guerin was considering signing defenseman Jon Merrill.
One of his first calls was to Sells.
“What do you think of Jon Merrill?”
“For how much?” Sells asked.
“850K,” Guerin said.
“That guy can play,” Sells said. “Jon Merrill at one-year, $850K, that’s a home run.”
Sells also suggested the Wild sign Merrill early on to a three-year extension (at $1.2 million AAV) a year later. Merrill isn’t a big-ticket player, but for that number for a third-pairing defenseman and penalty killer, it fit.
As of August, the Wild ranked No. 4 in the league in contract efficiency, according to The Athletic’s Dom Luszczyszyn’s model. Deals for Jared Spurgeon, Jonas Brodin, Ryan Hartman and Joel Eriksson Ek all have Sells’ fingerprints on them.
Guerin said he didn’t really pay attention to analytics during his player career — he didn’t believe in them because he “didn’t have to.” But he learned to appreciate them while an executive with the Penguins and brought it to Minnesota.
Advertisement
He bristles a bit when the Wild’s analytics staff is described as “lean.” He admitted there can be friction between scouting and analytics, but he doesn’t feel it here. And Sells said Wild scouting director Judd Brackett is very in-tune with analytics in building the league’s eighth-best prospect pool. AHL Iowa coach Tim Army said they have video coach Keith Paulsen create some of their own analytics data and get help from Sells, too.
“We have one guy who does our analytics,” Guerin said. “But it’s a big department because we highly value what Mat says, what he brings to the table. I don’t care if you have one guy or 12 guys, it’s the relationship of the people. Mat has a great relationship with me, the coaches. He speaks hockey. That’s why I think he’s effective.”
Guerin said he’ll add to staff when Sells feels it’s needed. And Sells believes they’re in a good place, considering they collaborate so much. “The staff may be lean,” Sells said. “But the investment (in analytics) is substantial.”
Sells usually sits next to Guerin and assistant GM Chris O’Hearn at games. He could do his work from his home in Edina, but the Wild feel it’s important for him to be there. Sells has his laptop with him at games but doesn’t use it unless there’s a question. His work starts after the game and early into the morning when he goes over data.
Sells loves using public data, citing HockeyViz, Natural Stat Trick and Evolving-Hockey as some of his sources. He digs Luszczyszyn’s models, too. “His work is fantastic,” Sells said. The public data isn’t typically in Sells’ reports, though there are times when a data visual is brought up in a meeting (like a heat map of shot attempts, or a player’s expected regression over a contract).
Most of the content in Sells’ reports comes from the mountain of data available to NHL teams through player and puck tracking. The puck has a chip in it, and there’s one in every player’s jersey, and that information has been a game-changer for teams. Public sites routinely use play-by-play data, for example. Sells said “probably less than 1 percent” of data teams use is available to the public.
“With player and puck tracking, we know where every player and the puck is multiple times per second,” he said. “Play-by-play tells you there’s a shot on goal from X feet. I don’t know how many data points there are in a game in player-puck tracking, maybe tens of thousands?
Advertisement
“The important stuff is now that you have this ocean of data, what can you do with it? How do you turn it into something that’s actionable?”
Sells wouldn’t get into many specifics on what he mines from player tracking data. He did say the team’s expected goals model is much more accurate. Like most teams, the Wild track scoring chances, 50-50 battles, opponents’ tendencies (their forecheck, their retrievals, where players are most dangerous on the ice). Public data offers outcomes, but internal metrics tell you “why those results are happening,” he said. For example, stats will tell you Brodin is strong at suppressing shots; puck tracking shows what he does to be so effective. Guerin often tells players that with the chip technology, everything you do on the ice is monitored and recorded. “You can’t hide.”
“All the best teams use statistics in every facet of what they’re doing,” Sells said. “Five years ago, every team didn’t have someone working in that area. Now you’re talking about teams that have 10-plus. Even for a while, it was ‘We’ll check a box.’ And if a reporter asks, ‘Hey what are you doing with analytics?’ they can say, ‘We’ve got a guy.’
“Now it’s part of everything.”
After practices or skates, Marcus Foligno finds time to do his homework.
The veteran winger will pull out an iPad or computer in the team lounge and watch his shifts. The information players have is endless. He can see his takeaways, his turnovers, where the puck travels while he’s on the ice, a heat map for each period and so on.
“It’s humbling,” he said.
But it’s also enlightening.
Foligno, on a shutdown line with Eriksson Ek and Jordan Greenway, noted that he’ll be happy if his line has more offensive zone time than time at the other end. Even if they don’t have a ton of scoring chances, they’re doing their job. The individual stats and video can be helpful, but what Foligno and teammates say is most useful is tendencies of opponents.
If Sells’ numbers point out that a team is last in the league in recovery on rims, the Wild will keep dumping the puck in. If the opponent is great at it, they’ll do a soft chip.
“That kind of stuff has been an eye-opener before games,” Foligno said. “Every team has similar strengths, but if you can find a weakness, you just try to beat the crap out of a weakness and put it in your favor. That’s what Sells and the coaching staff do a great job of.”
Foligno said analytics also help with mid-game adjustments. There might be a time when Evason will tell them they’re not getting enough second chances and chip tracking shows it’s because they’re not around the net enough. Foligno might use the technology to compare himself to a similar player and his wall play, techniques. For the penalty kill, the numbers will show where a power play typically shoots and scores from.
“Numbers don’t lie,” Foligno said. “If you’re doing well, numbers are going to be in your favor. The eye test is great. But if analytics can back it up, why not use it?”
Many #NHL teams have players wear heart rate monitors during practices, workouts, games. What do they learn from the data? An inside look how #mnwild use it to help determine how/when they practice, when players return from injuries @firstbeat https://t.co/HhR0fvkvzq pic.twitter.com/bgpyagUNlt
— Joe Smith (@JoeSmithNHL) January 12, 2023
It’s hard to find a decision Evason doesn’t examine the data for. He trusts strength and conditioning coach Matt Harder and the information gleaned from players’ heart rate monitors when it comes to how often the team practices. Sells’ analytics helps Evason with lineup decisions, pre-scout strategy and what to teach players in practice.
It’s not just reactionary, like the meeting the morning after the Oilers win. It’s preventative.
Evason said there are times when Sells might bring up a player who is trending in the wrong direction analytically so they can get ahead of it. The same goes for when the team is on a winning streak, but the process — the data — shows bad habits forming. If the coaching staff is concerned with a defenseman like Addison’s gaps, Sells has the stats to show he allowed six zone entries off his side in that last game.
“We do embrace (analytics),” Evason said. “I think it’s wonderful. But there’s a balance. You can’t have all analytics and you can’t have all hockey. You have to have a little bit of both.”
(Top photo of Mat Sells: Brandon McCauley / NHLI via Getty Images)
Source : From the Web