by Tracey Lindeman
Published in the Montreal Gazette on April 8, 2015
Ah, sports analytics. Reviled among traditionalists who believe it’s nothing more than soulless, cubicle-bound geeks typing furiously into calculators, but well-loved by people who believe stats measured by computers could hold the key to better, harder-working athletes.
Craig Buntin is in the second camp. He’s a former Olympian and the CEO of Sportlogiq, a Montreal-based startup looking to revolutionize the way players’ skills and performance are evaluated.
“There’s a common statement: Analytics are a lot like a drunk leaning up against a lamppost. They’re good for support, but they’re not really good for illumination. And this is specifically why people think it’s garbage — because it tells you what you already know,” Buntin said.
Buntin spent at least 20 of his 34 years on this planet as a competitive figure skater, but shifted into the business world after earning an MBA at McGill University in 2013.
“You are crazy. That’s impossible— I’m in!”
He founded Sportlogiq shortly thereafter with Mehrsan Javan, the holder of a PhD in computer vision — a field of study in which computers interpret movement and activity through images.
When Buntin first approached him with his idea, Javan said he told him, “You are crazy. That’s impossible — I’m in!”
Javan had developed a security surveillance tool whose underlying mechanics would eventually form the basis of Sportlogiq’s main offering: a single-camera, multiplayer motion analysis platform designed to understand and quantify athletes’ performance.
“So basically, intelligent game footage is what it is. We’re taking standard game footage from cellphones, from YouTube, from wherever, and we’re pulling stats out of it,” Buntin said.
It has the ability to compare players from across the National Hockey League, as well as the potential to be scaled down as a smartphone application-based metric tool for minor-league athletes and their parents. Eventually the company intends to broaden its scope to include basketball, soccer and other sports.
The idea of drunks leaning up against lampposts may be one way of describing sports analytics, but the team behind Sportlogiq insists measuring performance data by computer is the way scouting and tracking player development will be done in the future.
“If you’re scouting, 200 players a year come into the NHL from the CHL (and international and major junior teams). You’ve still got to go look at that game footage and subjectively say, ‘Yeah, I think that kid could be good.’ What we’re doing is potentially throwing thousands of players into a giant system where scouts can automatically say, ‘Well these players are developing well, these players have talent, we should keep an eye on these kids,’ ” Buntin said.
“It would provide opportunities for kids who might not otherwise get noticed. So you’re getting better talent, harder-working kids noticed and into the track on their way to the NHL.”
Sportlogiq has one NHL team on board so far and is looking at closing deals with two others before the draft at the end of June. So far, with a team of nine and a main product still in development, the Sportlogiq crew is able to turn around stats about 12 hours after a game wraps up.
So how do they do it?
“First of all, a lot of people lose sleep,” said vice-president of engineering Philippe Desaulniers.
They use computer vision, machine learning algorithms and big data analysis to track players on screen and capture a litany of game events including skating, passing, hitting, speed, dump-out success rate and at least another 130 or so game stats.
Before the visual data gets compiled into a numeric table and transformed into graphs, it gets checked by a so-called hockey expert — a blogger who for the past five years has obsessively tracked stats in his free time.
“Is he a fan or is he an expert? We’re looking at the data that he did even before he got here, and the guy’s an expert,” Buntin said.
That guy is Christopher Boucher, the man formerly behind Boucher Scouting, and he’s Sportlogiq’s newest acquisition.
“Numbers and hockey always went together for me,” he said. It began early, when he wanted to measure his own performance as a young goalie.
He freelanced as a stats analyst for some teams and was a finalist for a gig with the New Jersey Devils before finding a home at Sportlogiq, where his role is to review and correct or validate the computer’s analysis of plays.
“I’m excited about the fact that we can actually change the way players are described, the way players are analyzed, the way players are even scouted,” Boucher said.
It’s not about recognizing skills, Boucher continued, but what players do with those skills.
The goal is to eventually offer real-time analysis — something a coming $1-million investment round will help them accomplish. They intend to use the new money for product optimization and front-end development.
Next year, the NHL will begin putting microchips on players’ gear to compile its own raw data for presumably internal use only, but Sportlogiq is moving at a faster pace than the league. The company can already provide the kinds of data teams are seeking, Buntin said.
“So potentially, we might be the ones who start defining what the new language of hockey is,” he continued.
Javan interjected, correcting his co-founder. “We are the ones.”
Will Virtual Reality Be the Future of In-flight Entertainment?
Backroom bean counters move to the centre stage in times of crisis