Andrew C. Thomas Asks What Can Data Do for You
Thomas sees his puck and player tracking work at SMT as a continuation of his lifelong passion for solving problems with science and statistics.
Andrew Thomas only knew Eric Lander as his first biology professor when he started his undergraduate studies at the Massachusetts Institute of Technology in 2000.
Thomas later recognized his former professor as something of an intellectual role model and kindred spirit.
"He was a math guy from a young age, but he took his time figuring out what he wanted to do because he had so many broad interests," Thomas said of Lander, who graduated as Princeton's valedictorian at the age of 20 and needed only two years to earn a mathematics Ph.D. at Oxford before he shifted his focus to biology and became a central figure in genetic research, including the Human Genome Project. "He realized (with genetics) he had this area of interesting growth in the world with lots of good, challenging problems to get into."
Thomas did not find his calling in biology -- he ended up minoring in biology while majoring in physics at MIT. A doctorate in statistics from Harvard University and a visiting professor role at Carnegie Mellon University moved Thomas closer to what he wanted. Through sports analytics -- hockey analytics, specifically -- Thomas ultimately found a set of problems that called to him the way genetics lured Lander.
Thomas co-founded War On Ice in 2014 and watched it become a go-to resource for data-oriented hockey fans. He joined the Minnesota Wild as the team's Lead Hockey Researcher in 2016. And in 2019, he became the director of data science at SportsMEDIA Technology Corp, which collects the NHL's puck and player tracking data.
"I was always at my strongest when I was doing a bunch of different things or trying to do those bridging areas between subjects, but that didn’t lend itself as well to advancement in academia," Thomas said. "It’s a double-edged sword, but I don’t regret taking it the way I did because it’s what I enjoy the most.
"That’s half the reason I got into statistics in the end — it was another avenue to a lot of interesting pathways without necessarily choosing what I wanted to do with the rest of my life. There were lots of people doing lots of interesting things with a core skill set that was going to prove itself to be useful in lots of different places.”
“It was a real buffet of interesting problems to work on"
Well before Thomas began to develop a skill set that would unlock diverse work opportunities, he appreciated the possibilities created by career pivots. His parents certainly exhibited adaptability.
Thomas's father played bass in rock bands around Toronto. Around the time kids entered the picture, Thomas's father went to college and studied engineering. He later added a master's in business and built a career in supply chain management.
Thomas's mother worked as a chemist for the Ontario government at the time of Thomas's birth. She later set science aside to become a quilting teacher, a romance writer and an employee for Rogers Communications.
Until fairly late in high school, Thomas hoped his academic range would help him land a prestigious Canadian scholarship to study in England. Unfortunately, the scholarship was nixed as Thomas entered his senior year.
That reality raised the question of whether Thomas should remain in Canada for college or head to the United States. He hadn't given American academia much thought until that point.
“The first thing that really appealed to me at MIT was you didn’t have to declare a major until you had finished your first year of school," Thomas said. "If you go to a school in Ontario, you enter a program right away."
The progression from biology to physics to statistics to data science and from MIT to Harvard to Carnegie Mellon followed during the next decade-plus -- Thomas also worked as a professor at the University of Florida for one year between his time at Carnegie Mellon and his start with the Wild.
As a professor and advisor, he worked with students to develop models for player performance in sports, consumer behavior in marketing and carbon capture in environmental engineering.
“It was a real buffet of interesting problems to work on," Thomas said.
"I could’ve gone to the hedge fund business or worked in finance and done quite well financially, but I probably would’ve rotted from the inside, because I just didn’t enjoy anything I’d seen or heard about the culture or the problems that people were working on. After I finished my Ph.D., I got the same calls from people doing automated trading. All the recruiters were calling for Ph.Ds to do algorithm coding in some dark room for a long time with the promise there’d be a lot of money. It just never appealed to me more than being around people who were interested in doing things other than making money.”
"That’s why the download button for data was there from day one"
The idea of academia -- robust debate and rigorous work on a range of subjects -- eventually clashed with Thomas's reality. He questioned the findings and methodologies of some research that garnered attention in his area of expertise, networks.
"I was so fed up with that level of academic politics, and I was just having arguments rather than doing any good," Thomas said. "That’s kind of when sports came back to me as a viable option. I felt like at least in the sports world, I could maybe get someone to listen to me about something. The way that people like Eric (Tulsky) were doing, getting feedback from teams and getting them to pay attention...that appealed to me because I always wanted to communicate with people about the things that I thought were interesting.”
Thomas's serious interest in sports analytics traced back to 2003, when he read Michael Lewis's "Moneyball." He sent emails to several pro baseball and hockey organizations before graduating from MIT in 2004 just to see if any might take a chance on someone with his skills who also shot photos of the Boston Red Sox for the student newspaper in his spare time.
While at Harvard, and well before he came across hockey analytics blogs, Thomas started manually tracking zone exits and entries for research he published in an academic journal in 2006. He also dug into papers written by Brian Macdonald and Alan Ryder on hockey and Ben Alamar on basketball.
By the time Thomas met Tulsky -- now the Carolina Hurricanes' Vice President of Hockey Management and Strategy -- at MIT's Sloan Sports Analytics Conference in 2013, his views on the best way to publicize hockey research and engage the academic community had evolved.
“The very first reaction I got from those experiences was hey, people read blogs more than they read academic papers, so what can we do about that?" Thomas said. "I’d been completely on the other side of this. It was other people, like Sam (Ventura), who had been reading blogs and made me aware of this stuff.”
Thomas and Ventura -- now the Pittsburgh Penguins' Director of Hockey Research -- tackled sports research projects at Carnegie Mellon for more than three years before they committed to creating War On Ice.
Alexandra Mandrycky -- now the Seattle Kraken's Director of Hockey Strategy and Research -- added yet another star to War On Ice's team when she emerged as someone who could quickly add salary cap analysis to the site.
Thomas, Ventura and Mandrycky's careers certainly shaped the legacy of War On Ice. But Thomas hopes the hirings never become the defining detail of the site's history.
"We wanted it to be a place where cool research existed, and by the way, here are some other stats on the side," Thomas said. "The fact that the site was so useful to others was partly a function of us thinking about it as academics. We knew we wanted to share whatever results we had. We wanted to make sure people could get them and use that for whatever work they wanted to do. That’s why the download button for data was there from day one. It was like, 'Here, grab whatever this table is, put it in whatever you’re doing, and just cite us.'"
"It’s the fact that those communication lines are more open now than they were that’s fascinating to me"
Thomas thought he might utilize real-time puck and player tracking data soon after he accepted the job with Minnesota in January of 2016. After all, the NHL had hinted at it during the 2015 All-Star event and brought it up again as part of the 2016 World Cup of Hockey.
Years passed. The league-wide tracking data never came while he worked for Minnesota. Thomas made sure his next stop, SMT, put him in position to see the goods.
"I like to be a bit of a perfectionist," Thomas said, "and I like to collect all of the data I possibly can before making a lot of these judgments, which is probably why we've paused as long on a lot of the tracking data as we have. I want to be very clear and very sure that we’re picking up every explanation for some of these things. These things we’re measuring now like defense involve tracking players a particular way, and I don’t want to jump to any early conclusions with this stuff.”
While the treasure trove of tracking data eluded him during his time with the Wild, Thomas still found plenty of problem solving opportunities.
"When we sat down and talked about the WAR stuff, it was very clear that they were open to it," Thomas said of the Minnesota Wild executives that interviewed him, including then-general manager Chuck Fletcher. "They saw value in thinking about the kinds of things that we did while being open to the idea that we wanted to develop new things off of these same principles.
“A lot of what I was doing and how I evolved with the job was, with the data we had, how do you have conversations with people who see the game a different way so that you can prove to each other that you have something to teach each other and become a better team as you go along? Whatever data you have access to, that’s always a possibility, even with most basic stuff. You can convince someone that plus-minus is garbage because it’s too short of a sample size or it’s too dependent on the goaltender or because you’re on the fourth line and it’s not going to be as good of a measure. If you can get people to buy into that and then tell you something back about how they see the game, you will become better at what you do. … You will find a way to make a difference. It might not be as big as you like, but there’s something there that you can dig into.
"The greatest fear I had was being backroomed, being someone they stow away and don’t listen to.
“I’m very satisfied with the work I’m doing with SMT and being at the vanguard of where this new data is coming in. I’m very much enjoying the people I work with, the work I’m doing, and that I’m getting to do it first."
SMT's infrared-based tracking system actually allows Thomas to draw from knowledge he accumulated while doing his senior thesis at MIT. His research contributed to work done at the Laser Interferometer Gravitational-Wave Observatory (LIGO), which, among other things, detects new black holes in the universe.
"Finding these new connections in these pieces of data and bringing up all of the stuff I’ve done before, that’s the stuff that really animates me," Thomas said. "Even the stuff I did with my own play tracking back in 2005, it was all about zone entries and puck possession, things like that. I’m thinking about that now with the work I’m doing. The approach I was using 15 years ago when I was tracking games in Bright Arena at Harvard, was the way I was thinking about the game then still useful now? Having new data, what was I missing before? That really drives me.
"The most interesting thing to me about the new data coming in, tracking data in particular, is reaching out to hockey people and saying ‘I have this idea that I think will be helpful with the data. How would you look at the problem? And can we, together, come up with a way to identify this property that you know you can sell to a player or coaches and that I can find with data?'
"It’s that whole throughline that matters. It’s not just that I think I can come up with an interesting way to measure something. Take player WAR, for example. If there’s no way for me to connect the numbers I’m seeing to something that a coach can do or a GM can appreciate in deciding why they should sign a player or trade for that player, then it’s less valuable. Tracking data and even biometric data are offering new ways to talk to people that could not have been done before. Whatever that data says, it’s the fact that those communication lines are more open now than they were that’s fascinating to me.”