How AI is creeping its way into NHL front offices…
Hockey’s analytics motion lags behind other sports activities, but its early success tales are already nicely told.
There is the video of Canadiens scouting director Eric Crawford efficiently pushing to choose Lane Hutson in the second spherical, telling the room that every contending staff has a defenseman like him.
There is then-director of analytics and not too long ago employed basic supervisor of the Devils Sunny Mehta’s draft model that ranked Jesper Bratt third general in the 2016 draft, main Mehta to push to choose him in the sixth spherical.
There are figures like Hurricanes GM Eric Tulsky, Avalanche director of analytics Arik Parnass and Carolina assistant GM Tyler Dellow — former bloggers and writers who rose to prominence and helped construct contending NHL groups.
There’s still a hole in hockey between groups that embrace data and those that don’t, but the beginnings of the next wave of innovation have already began rippling. This one will probably be about artificial intelligence, and which groups can combine it into their operations best.
Eric Tulsky of the Carolina Hurricanes arrives for the sport against the Florida Panthers prior to Game Three of the Eastern Conference Final of the 2025 Stanley Cup Playoffs at Amerant Bank Arena on May 24, 2025 in Sunrise, Florida. Getty Images
The challenges exist on two ranges. First, in the game slowest to embrace analytics, how do you promote the leap to AI? Just like analytics, that wants to be answered with outcomes that show those who don’t adapt will probably be left behind.
The second is broader.
“That is a question that every company in America is asking themselves, and it’s a daunting one: It’s where do we start?” one Eastern Conference government told The Post. “Ultimately what it’s gonna be is every operational process will be rethought through the lens of AI.”
Analytics is a broad matter that will get pigeonholed into a simple one, but as it relates to any sport, you may boil down its function into a single thought: utilizing numbers, trends and data to perceive what the attention misses.
AI is not like that. If you had to boil down its potential use instances in hockey into one thought, it might be that it might save time. But that is so broad it’s barely a useful descriptor.
“The examples of how to use AI, I think the simple ones are … it’s useful for cap management and modeling, to analyze other clubs and other GM tendencies,” Steve Werier, a former assistant basic supervisor for the Panthers who’s since labored as an AI lawyer for Amazon, told The Post. “I feel it might help you establish indicators in other media markets that may be related to strategic planning.
“Another example is, if you’re a hockey ops executive and you get 100 agent emails a week about players in various leagues around the world. An AI agent with the right prompts can sort through those and highlight ones that most closely fit the criteria your minor league clubs are looking for and that saves you time. AI can allow analytics teams to speed up coding and up the execution timeline.”
What else? Teams usually make use of a handful of participant development coaches to meet with prospects in faculty or juniors around the continent during the season. What if, to supplement the time in between those conferences, you would automate AI to give prospects real-time suggestions?
Teams also use firms like SportLogiq, which has access to real-time monitoring data, to help construct inner fashions and give insights to workers. But might you employ AI to assist a coach midgame by accelerating how shortly data and video influencing matchup selections attain the bench?
“What if I were to give a coach an iPad that had real-time insights that said, ‘Oh s–t, your third line’s getting caved in?’ You may be seeing it anyway, but real-time adjustments on the fly to help him win games in the moment,” the Eastern Conference government said. “All of this stuff is possible, and it seems far out, but it’s not in the context of what is possible with the technology we have today.”
Mike Kelly, the director of analytics and insights at SportLogiq, homed in on participant acquisition as an space where AI can help.
“You’re rarely if ever saying let’s look at all the unrestricted free agents or even the restricted free agents, grade them out from best to worst and let’s just start plugging away at the best. You’re looking for fits,” Kelly told The Post. “You’re saying that perhaps we need a depth centerman who can win faceoffs and kill penalties. That’s what you’re trying for. Or we’re trying for a bottom-pair defenseman who’s bodily and kills penalties.
Assistant Coach of the Minnesota Wild Brett McLean reviews a play on an iPad with Brandon Duhaime #21 during a first period timeout against the Philadelphia Flyers at the Wells Fargo Center on March 3, 2022. NHLI via Getty Images
“You can input the data and have your AI tools, having certain contract information, maybe you want a guy with one or two years left tops or an unrestricted free agent at a certain price range, have it populate a list for you. Drill down once you’ve got your list into an even shorter list. That’s kinda the manual work that would have to be done. The AI tools can do that work quicker. So it’s a time-saver.”
There’s an apparent counterpoint to all of this that everybody who’s ever casually used an AI assistant can assume of: Isn’t its reliability suspect?
The reply to that relies upon on the context — how good are you at prompting it, how good is the data you’re giving it to work with and are you asking the correct questions about what it outputs? What everybody stresses, though, is how fast the technology is bettering, and how a lot time it might already save.
“A year ago, I was hesitant to use it because it may stray you on a path,” a Western Conference government told The Post. “I used to be studying my own way of it. And now it’s more generally used this season for sure. I wouldn’t say it’s used for on a regular basis, all-day form of factor. It’s more of a assist software to help pace up or work through issues, different issues like that.
“It hasn’t taken over and that’s where we’re at that precipice of, where is it gonna go and how are data providers gonna use it and how do we use it?”
Jules Lanari-Collard, a pupil at Imperial College London, not too long ago offered a model he created to measure how gamers contribute away from the puck at the HALO convention in Denver, a first-of-its-kind hockey analytics convention hosted by Parnass. He used Google’s Gemini to write about 10 p.c of the code for it, which in the world of statistics is considerably on the conservative aspect.
“For writing code, for making things more efficient, it can really speed up a lot of things. That’s one aspect,” Lanari-Collard said. “I think fundamentally you still need really good people. Those people may be able to implement AI models to develop metrics and things like that, but you definitely can’t just sort of hand off analytics to an AI and expect to get particularly good results.”
Mike Sullivan (L) and assistant coach Mike Vellucci of the Pittsburgh Penguins look at the iPad against the New Jersey Devils during the second period at Prudential Center on April 11, 2021 in Newark, New Jersey. NHLI via Getty Images
AI will finally be built-in into every group, but, rightly used, it gained’t exchange data analysts, scouts or anybody else. It will supplement their work and save them time.
If there’s one draw back, it’s that most of that innovation, at least in hockey, will occur away from the public eye. Unlike, for instance, baseball — where granular ranges of monitoring data can be found to anybody who desires to use it via Baseball Savant — hockey’s monitoring data is largely inaccessible unless you’re employed for a staff, the NHL or another company with access to SportLogiq. To do his project, Lanari-Collard and others who participated in HALO’s Hackathon got some of that data, but the average fan can’t access it.
“You want to break into baseball, American football, basketball, it’s really, really saturated,” Lanari-Collard said. “Not to say there’s not more that can still be done, but there’s a lot of space for innovation in hockey. We’re limited by the data available. That’s the bottleneck.”
What that might also imply, though, is a more noticeable hole between groups that embrace new methods of pondering and those that keep to custom.
On the staff aspect, though, that limitation doesn’t exist, at least not anyplace close to the same way as it does for the public. It’s just a query of who can innovate quickest.
“I think, in terms of hockey operations, there’s not going to be many places [AI’s] not going to touch,” the Western Conference government said. “… I can’t think of a department who wouldn’t benefit from the ability of it being used at some point in the next handful of years.”
Stay up to date with the latest trending topics! Visit our web site daily for the freshest Sports news and content, rigorously curated to keep you informed.
