The Big Read: Revamped analytics department bridges disconnect between data, hockey ops
The hirings of Matt Perri and Lee Stempniak reflect Alex Meruelo's continued investment in the team despite an assortment of COVID-related financial challenges
Welcome to the AZ Coyotes Insider newsletter. I generally publish stories four to six times per week (some of them free). By subscribing, you’ll be supporting independent, accountable journalism. Subscribe now so you won’t miss a story.
A few stories from Bill Armstrong’s past highlight the disconnect that sometimes exists between analytics and the eyeballs of NHL managers, coaches, scouts and players.
Armstrong was reading a comparison of Pierre-Luc Dubois and Patrik Laine in their draft year (2016). The data folks had Dubois light years ahead of Laine in their evaluation. The problem was the samples that they used for evaluation. Dubois’s 11-game sample came from a run of games in the QMJHL against 16-, 17- and 18-year old players — with many of those games played against the Q’s worst teams. Laine’s sample came from the 2016 World Championship where he was playing against men and still had seven goals and 12 points in 10 games.
Another analysis pushed across Armstrong’s desk suggested that Kailer Yamamoto and Ryan O’Reilly were comparables. Yamamoto is a speedy, 5-feet-8 winger. O’Reilly is a 6-feet-1, 200-foot center with a Stanley Cup and a Conn Smythe Trophy on his résumé.
Other reports that Armstrong has seen just have no way of quantifying or identifying character.
“You might have a situation where your analytics guy goes through the waiver list and says, ‘You should claim this guy,’ and you look at him and say, ‘You don’t understand. ‘That guy’s a dirtbag,” Armstrong said. “In an instant, your analytics guy has lost credibility with the entire staff.”
Armstrong believes wholeheartedly in the value of analytics. That’s why he just hired Matt Perri as the Coyotes new director of analytics. Perri helped build Sportlogiq into one of sports’ premier data sites as its senior director of hockey and pro services. His expertise will aid multiple levels of the organization, including management, coaches, players, scouts and trainers.
But Armstrong wanted to give Perri and his staff a conduit to hockey operations, one who could translate complex data into hockey language while filtering out information that had no application, or was simply off base. In addition to Perri, the Coyotes have hired former player Lee Stempniak as their hockey data strategist. Stempniak played 14 seasons for 10 teams (including the Coyotes) before retiring in 2019.
“In every position that we have looked to hire, Bill has said the same thing: ‘How can we be elite? How can we move the needle and find people who can bring something to the table that makes us better?’ So in evaluating analytics, we did a deep dive on what we needed,” said special assistant to the GM Brian Daccord, who was charged with identifying candidates for both positions.
“We interviewed every one of the department directors on our staff and asked, ‘What do you need from analytics? What are your expectations to make you a better scout, to make you better in player development, or to make you better in coaching?’ It all came back to usable information, not just pages and pages of stats but what can you actually use?”
As Armstrong and Daccord searched for possible solutions, they found one in baseball, where teams use ex-players as conduits between analytics and the players. The Diamondbacks do it with pitching strategist Dan Haren and run production coordinator Drew Hedman.
“It’s basically somebody who speaks the language of baseball,” Daccord said. “So when the analytics department says a pitcher should be throwing more curveballs, instead of a data analytics guy who never got past minor league ball talking to a pitcher that makes $20 million a year, you get it from a former pitcher. It’s a little hard for that pitcher to accept it from a guy who hasn’t played at this level but if it's coming from a former big leaguer he is able to frame it in a way that the pitcher is able to use; athlete talking to athlete.
“Bill and I both loved that concept so we set out to find a guy who was incredibly well respected in hockey and has a ton of experience, but is a really intelligent, curious, inquisitive guy; somebody academically inclined.”
Stempniak was in that transition period between a playing career that he had to accept was over, and what was next. His last full season came with Carolina in 2016-17. He injured his back at training camp the following season, came back at Thanksgiving on a conditioning stint and broke his collarbone “about five seconds into my first shift,” he said. “I didn’t play my first game until mid-January.”
Stempniak didn’t want his career to end that way so he reported to Bruins training camp on a tryout in 2018, spent most of the year with Providence in the AHL and played just two games in Boston before admitting the end had come.
For the past year and half, he has been coaching his kids, spending time with family and volunteering as a high school coach, but he wanted to get back into the NHL. He sought advice and information from NHL executives such as Minnesota Wild GM Bill Guerin. When Daccord reached out, it felt like a perfect fit for both sides.
“My eighth grade teacher pulled my parents aside at graduation and told them I should pursue a career in math in college,” Stempniak said, laughing. “I guess I filed that away for 25 years, but I was an economics major at Dartmouth so numbers were always something that interested me.”
Stempniak, 37, was still playing when analytics such as Corsi first hit the mainstream. He knew coaches such as Carolina’s Bill Peters and Arizona’s Dave Tippett were using data to help their team, but it was more a point of interest than applicable knowledge for most players. He believes the time has come to alter that attitude.
“Analytics are here to stay and they have come such a long way from when I was a player,” he said. “In Calgary (2011-13), we talked about shots and where they came from but now there is so much information out there that it has the potential to be really impactful.
“The main thing for me is that it’s objective. You can't have hard feelings when a coach shows you a number because there is no bias to it. As a player, you like accountability and you like honest answers from the coach, but when you hear, ‘Hey, you’re not playing well’ or ‘You have to try a little bit harder,’ that’s the worst thing you can hear from a coach. But if you hear, ‘You need to make better plays from the boards’ or ‘You're not generating enough from these areas,’ you know what to work on. That’s the value of analytics.”
It’s not as if Perri, 33, is some ivory-tower academic sitting in a lab, cut off from the hockey world. He was born in Edmonton and he played the game growing up. Once he gave up the life of a musician and started working for Chris Boucher (now with the San Jose Sharks) at Sportlogiq six years ago, he worked closely with teams so he understands their language and concerns.
“I think this position is largely a transfer of what I was doing for Sportlogiq,” he said. “A lot of teams were using Sportlogiq insights, not just data. Chris and I were part-time analysts for a lot of teams and we got exposed to a few different techniques so what I will do for the Coyotes carries over.”
One of Armstrong’s mantras since he took over as GM is to make the Coyotes a purposeful and efficient organization. Perri thinks he can help.
“One of the advantages I had working for Sportlogiq is that I saw how a lot of teams did things and I saw a lot of the inefficiencies,” he said. “Analysts were spending more than half the time doing raw manual data collection or cleaning, but if you have analysts, you want them to be analyzing.
“What I really want is to build a process where it's less about a scout calling a GM and saying, ‘What do you think about this player?’ and more a distinct view on things where everyone in the department follows the same process and procedures. I want to automate a lot of these processes so we don't spend a lot of time on it. I hope to build Bill’s trust and the scouts’ and coaches' trust in the work I am doing because I think there are more instances in which this data can be used.”
Practically, Perri said he expects to create pre-game and postgame reports, as well as bigger-picture reports that encompass 5-10 game segments or serve as midseason reports. The pre-game reports will cover everything from an opponent’s style of play, to how they match up, how lines have been performing recently, special teams, goaltending and tendencies. Postgame reports will identify issues and areas of success.
The area where Perri sees the most potential for growth is draft data.
“We have a lot of data on amateur players and this is one of the areas where teams are most concerned,” he said. “There are so many players to look at and people to coordinate and I think data can provide a huge advantage, but it's still a new avenue.”
Like many analysts, Perri groaned when the NHL pulled its microchipped pucks from the ice this season after a handful of games, due to performance issues. He knew that the data would take time to comprehend and contextualize, but he would love to see more high-end data hit the mainstream for consumption by reporters and the general public. Once that is achieved, he said it can help solve an existing problem: a lack of quality data.
“There are analysts, people who produce blogs or actual analytics writers in the public who are squeezing everything they possibly can out of limited data and they've got very good models, sophisticated approaches and they are doing everything right with the data that they have but they just don’t have access to this high quality data that even a GM or an assistant GM of an AHL team has,” said Perri, noting that Sportlogiq produced data from between 3,000 and 4,000 events per game.
“Those GMs don’t necessarily have to have any of these advanced approaches and models to look at the data. They can just count things like the percentage of times a player is in a D-zone, or scoring chances for and against. There is such a granularity of data that you don’t need fancy approaches.”
Perri and his staff will work closely with Stempniak to produce digestible, usable, succinct reports that recognize that their audience has massive time constraints, whether it’s Armstrong or the coaching staff.
“For me this position is about building a bridge in the translation of this data and also filtering out what doesn’t make sense,” Stempniak said.
While the perception still exists that some GMs and coaches are averse to analytics, Stempniak believes the hockey ops people have a thirst for such knowledge, as long as it is presented properly.
“I think they view it as another source of information, another tool,” he said. “The margins are so small between winning and losing in the NHL that any small advantage you get helps, so if it helps them win and helps players play well, they’re all for it.
“Look at this season where you play the same team seven, eight, nine times. Within that structure you can unearth the things that make teams successful, but at the end of the day, it has to be in conjunction with other things. There are some things that analytics will just never capture like hockey sense and competitiveness. You can measure who wins puck battles and quickness but you can’t always measure what drives guys.”
What the Coyotes new analytics department can achieve is still unclear. All parties understand that it will evolve over time.
“Bill will be the first person to tell you this isn’t turnkey,” Daccord said. “This is a new way of doing things, a new approach and it will take time to create chemistry.”
As hire after hire settles into their new role, however, Armstrong said it is clear that owner Alex Meruelo is willing to invest in the organization despite a difficult financial year, brought on by COVID-19.
“That's one of things they talked about when I was coming in and applying for the job,” Armstrong said. “I asked if I would have room to hire and they said, ‘Absolutely.’
“What they have said in the meetings about what I could do and who I could hire has been exactly as advertised. They have backed me 100 percent financially in the middle of a pandemic to hire the right people. We’re not all the way there but we’re getting there.”
Follow Craig Morgan on Twitter: @CraigSMorgan