My introduction to hockey analytics came in the offseason following the 2011-2012 season. With the lockout looming, it was obvious that there would be a lot of free time to read about all things hockey. A person can only take so many “culture change” or “best shape of his career” articles before questioning the purpose of life, though.
I don’t even remember how I really got into it. It was one blog after another, after another, after another. All of a sudden, I was down a wormhole of blogs from noted #fancystats writers like Eric Tulsky (who was recently hired by an unnamed team, for privacy reasons), Tyler Dellow (who now has a regular column on the Sportsnet and has frequent spots on Canadian national sports radio), Gabe Desjardins (who also consults for pro teams and runs the BehindTheNet website), and Corey Sznajder (who is tracking zone entries/exits for every single NHL game from last season by hand). I’m forgetting many smart people, those are just the ones that come off the top of my head.
It was a time when analytics were being talked about more on the blogosphere and Twitter, yet hadn’t really gained much public traction. In fact, even through this past season, several notable sports people (in Canada at least) would publicly denounce the use of things like Corsi, Fenwick, Zone Starts, Quality of Competition/Teammate and so forth. And I do mean publicly:
https://twitter.com/simmonssteve/status/395375707821309952
https://twitter.com/DamoSpin/status/432329868370378753
The first is a writer for the Toronto Sun, the second for the Toronto Star, the third is a national talking head on Canada’s The Sports Network (TSN). There are more, but readers get the idea.
Perception of #fancystats (the hashtag has become a Bat Signal for those that both endorse and detest analytics in hockey) is changing. Personally, I use them as part of my fantasy hockey research for what will be the third season now coming up. There have also been NHL teams speaking directly about such things: Ron Hextall (the new Philadelphia Flyers general manager) said the Flyers will be taking a more analytical approach to decision-making; it’s no coincidence that two former Toronto Maple Leafs in Nikolai Kulemin and Mikhail Grabovski signed with the New York Islanders. Those two guys were a 50.9-percent CorsiFor together from 2010-2013 with the Leafs when Toronto was (is) an absolute black hole for possession (only Edmonton and Minnesota were worse as a team over that stretch); St .Louis Blues head coach Ken Hitchcock has talked publicly about his and other teams’ use of analytics in recent interviews.
If the use of analytics in hockey wasn’t apparent or embraced by hockey fans before this week, it certainly should be by now. On Tuesday, the aforementioned Maple Leafs – who, mind you, have let possession players like Clarke MacArthur walk, bought out Grabovski, and gave David Clarkson a bigger cap hit for more years than John Tavares or James Neal – hired former Sault. Ste. Marie Greyhounds general manager Kyle Dubas to be an assistant general manager to current Leafs GM Dave Nonis.
If the name isn’t familiar, a little background on Dubas. The 28-year-old Dubas just finished his third season with the Greyhounds of the Ontario Hockey League. In this great piece from TSN’s Scott Cullen, Dubas noted that the team’s possession ratings (with guidance from head coach Sheldon Keefe) rose from 47-percent to 57-percent in their tenure. With the use of hand-tracked possession metrics like Corsi, Zone Starts, exits/entries, the team went from last in their division in Year 1 to winning the division in Year 3. A turn-around, to be sure.
The same is hopeful for the Maple Leafs. The team hasn’t made the playoffs in an 82-game season since I was in high school (my 10-year reunion is this coming winter). The only year they did (2013), it’s probably because the team didn’t have time to collapse in the 48-game season, as they did in 2012 and 2014. This team needed a fresh approach to the game, and Dubas will certainly be able to help them not only do that, but build a sufficient analytics department within the organization as well.
Let’s be clear here. There is no “magic bullet” statistic that so many cavemen seem to assert is expected. There is no one statistic that says “this player is bad, this player is good.” Everything is given in context. A player had a bad possession rating, but what was his competition (first or fourth lines)? Who were his linemates? What were his zone starts? Was he playing injured? Was he on his off-wing? The questions are endless, and there is context needed for every analytic measurement used.
What this does signal that the junk of “leave hockey to the experts” or “CORSI Hockey League” is over. Most of these analytics are available on websites like BehindTheNet, Extra Skater, ShiftChart, and Some Kind Of Ninja. The analytics “revolution” is over, but there is still a lot of work ahead. The NHL will begin testing SportVu, which is already in basketball arenas. It’s a technology that will help optimize player positioning both defensively and offensively, among other things. There are still discussions to be had about how to get the puck to good shooting areas, how to create a play to get the goalie moving, what’s the best power play setup for a given set of players, and so on.
All that said, it’s amazing how far analytics in hockey have come:
https://twitter.com/mc79hockey/status/491625079152074752
Possession metrics started as a bunch of guys on a hockey forum. Now it’s to the point where an assistant general manager for the most valuable franchise in hockey is a 28-year-old who embraces the use of analytics as part of the decision-making process.
The debate is over but the discussion has only begun.
