Sports News Without Access, Favor, Or Discretion

Player-Tracking Technology Is Coming To Hockey. It'll Change Everything.

Nothing happens quickly in the NHL. Fighting still exists. Visors were only mandated for incoming players last year. But the league is reportedly preparing to test player-tracking technology, and if it takes, we're not far away from a complete and sudden revolution in how we understand the sport.

You know all about the general skepticism of the kind of "analytics" — if you want to call shot attempt plus-minus analytics, which you shouldn't — that people have been putting together to evaluate teams and players alike in recent years. The argument against them has shifted from, "They don't mean anything," to, "They don't tell you what you didn't already know," and that is on some level a failing of the numbers themselves. The stats corsi and fenwick get thrown around a lot these days, and they tell you quite a bit about both process and result; The more shot attempts you have, the more likely you are to score goals, and the more likely you are to win. That's simple stuff. But they're also relatively new and thus they are often viewed as bad, sight unseen, for one simple reason.


What about what happens in the time between shot attempts, the things that don't come out in "the stats?" Well, they do, kind of, but the numbers that are currently used are widely regarded as being terrible. The NHL officially tracks "real-time stats" (hits, takeaways, and giveaways) to try to fill in the gaps, but these are cursory at best and spurious in all cases. Teams count them differently, and don't seem to follow anything resembling the NHL's official definitions. The Islanders' Matt Martin, for instance, led the league in hits this year with 359, nearly a hit-per-game ahead of second-place Cody Franson. Two more Islanders were also in the top 20, so this tells you two things: The Islanders didn't have the puck very much, and they probably overcount hits. What it doesn't give you is an analysis of Matt Martin that can be used to compare him to other players in any meaningful way.

Fortunately, it seems as if the league is finally trying to change that. It was reported over the weekend that the NHL is looking at using the kind of player-tracking technology already long-since put into use by the NBA as a means of pulling so much more data out of the game than before, and filling in all those little gaps between shot attempts. It could be in place for the start of the 2015-16 season.

Here's what the visualization of the SportVU system looks like for basketball, as it tracks the speed, movement, and possession of every player on the court.

A potential problem with applying the same technology to hockey is that much more complex than basketball. No one's sure if the SportVU technology, which will use cameras overhead and along the glass to track every inch of the ice, can deal with three things that happen a lot in hockey: Line changes, sticks swinging around wildly, and the puck doing wacky things. Basketball has a lot of chaos within the game, but people aren't coming on and off the court at the same time, and the ball is almost never under people, or flying out of play at high speeds.


Tracking the puck is a major challenge for SportVU cameras. To that end, Sportvision, a competing system — and the company that pioneered the infamous "Glowpuck" on Fox all those years ago — is also in the discussion with plans to put another (presumably much smaller) chip in the puck to track its movements. Even if it's not there quite yet, this technology is one genie that won't be put back in the bottle. However it's eventually implemented, player-tracking is going to be a part of hockey before long.

Analyst Eric Tulsky, who has worked with NHL teams to develop better ways to measure the game, says the expectation is that most teams would use this kind of technology only to do the simplest things at first: Track and perhaps standardize what constitutes a scoring chance. The league doesn't officially do this, and though many bloggers do, there is no uniform definition. Plus, teams already have this stuff for themselves, internally, and in general the people who have tracked this kind of thing for years find that there's a strong correlation between carrying the puck into the zone cleanly and producing more shot attempts than if you just dump it. Basic stuff, really.


Tulsky also said that those teams which are "ahead of the curve" when standardized video tracking hits the scene will probably use the data these systems provide to track more than the basics. Things like those zone entries and exits, and how that translates to higher-quality scoring chances will all be quickly counted and put to use. This, too, is currently being tracked manually, from bloggers and teams alike watching the games and compiling the data into something intelligible. Watch dozens of these per game, over the course of a season, and patterns emerge.


There's value there, and that's why some teams invest heavily in these stats: Spending a few hundred grand to get a handful of extra points out of the season, or saving that much in free agency by signing guys whose other stats might not be reflective of their true talent, is exactly the kind of exploitation of inefficiency that built the statistical revolutions in baseball and basketball. It's an advantage only because not everybody is using it.

But if these systems can actually work well in-game, then the entire sport is going to change more or less overnight (in hockey terms, that means "over a few years"). They'll go from tracking simple things like shot attempts to very advanced things like this in short order. While it won't make it onto league broadcasts, fans who really want to dig into it might be able to look at what an idealized "Ghost" Sidney Crosby did and see how that compares to Real Sidney Crosby. Is he actually missing a step in this Rangers series, perhaps due to injury? A tracking system could quantify that down to as many decimal places as you need.


The big question, though, is whether the NHL is going to make this data available to the public. If it's not being run by teams, but rather centrally by the league itself, then there's no reason not to. But because this is the NHL, it's tough not to be doubtful. You can't find corsi and fenwick, or even zone starts, on Some league employees have pushed for it, but it hasn't happened yet. It's another impediment to the widespread acceptance of these stats as being relevant. Compare that to the NBA, where its SportVU data is right there for the perusing.


Perhaps basketball's not the right parallel. A straight line can be drawn between the way this would theoretically work in hockey and the way it currently does in soccer. Major sites and even television broadcasts talk about the rates at which a team or individual completes pass attempts, how much distance players have covered in games, the number of shots that are generated from individual players' passes, etc. This data is easily quantifiable for hockey, if somebody took the time to track it. They already probably do, somewhere. But again, hockey doesn't like to talk about this kind of thing. Bruins broadcasts talk about zone time, as though corsi weren't a more or less accurate representation thereof, and the team itself weren't investing in more advanced analytics. Penguins broadcasts talk about corsi with regularity, though usually (and oddly) not by name. So progress is being made, but slowly and with considerable resistance. This data has to be uniform and easily accessible for most people to really accept it, and not write it off as more pointless numbers for nerds.

The more we can glean from these systems, the better we'll understand the sport. That would be a big change on its own, because most teams right now just think they understand it. Take the Toronto Maple Leafs, who famously eschew all attempts to evaluate their team other than Randy Carlyle's backwards say-so, and have as a consequence been pretty awful the last two seasons (and likely will be for the near future, given that they recently extended the coach for another two years).


Maybe they'll learn something with SportVU or Sportvision. Maybe they won't. But if this data movement works, and if other sports are any indication it does, then Toronto and teams like them will be left even farther behind than they already are.

Ryan Lambert is a columnist for Puck Daddy, among other places. His email is here and his Twitter is here.


Top image via

Share This Story