Fancy Stats Nerds Gather to Discuss Hockey Analytics’ Future
On February 7th, some of the biggest names in hockey analytics came together in Ottawa at Carleton University.
Dubbed #OTTanalytics, the conference represents another step for “fancy stats” towards the mainstream. Something that old-school blusterers like Don Cherry and Brian Burke are sure to hate. Hockey Abstract’s Rob Vollman said it best in his opening remarks.
Referring to the analytics event he spoke at in Edmonton, Vollman noted that they used to get 20 people to an analytics conference. For the weekend’s conference at Carleton, the organizers had to cap registration at 225.
And as War-On-Ice’s Sam Ventura tweeted:
Analytics is really starting to boom.
The days of clamouring for dump-and-chase, more fights, and careful attention to your favourite player’s plus-minus ratings are long gone. In its place, hockey nerds and fancy stats have taken over.
Heavy Hitters: The #OTTanalytics Conference Presenters
The Ottawa Hockey Analytics (#OTTanalytics) event featured an all-star lineup of current thinkers leading the math nerd movement in hockey.
Ranging from the “old guys” (Vollman’s term) to the new guys, speakers at #OTTanalytics included;
Vollman – Hockey Abstract; David Johnson – Hockey Analysis;
Andrew Thomas and Ventura – War-On-Ice;
Tom Awad and Timo Seppa – Hockey Prospectus;
Josh Weissbock – CHLStats.com; Stephen Burtch – Sportsnet;
Andrew Berkshire – Eyes On the Prize; Michael Schuckers – Statistical Sports Consulting;
And new fancy stats voices Matt Cane (puckplusplus.com), Emmanuel Perry (Senstats.com), and Alex Diaz (alexdiaz.ca).
Let’s take a quick look at three themes that dominated across speakers.
Fancy Stats Summary
1. Corsi is Dead
Ken Campbell at The Hockey News stirred the advanced stats world with his article, “The death of Corsi? It’s only a matter of time now” on January 24th.
Though Campbell is far off the mark in proclaiming the end of stat matrices that seem to trouble only “old white guys,” the #OTTanalytics presenters agreed with the death of Corsi…sort of.
Many of the presenters (Ventura, Cane, Awad, probably others while I was distracted by live tweeting the event) noted that accurate data on shot types and shot quality are key to developing new analytics that better measure player performance.
Sam Ventura promoted the use of “expected goals for.” He explained that, to predict future player performance, data on shot type (backhand, wrist, etc…) and shot quality (shot from the point or rebound in the slot) must be added to Corsi data.
Cane spoke most clearly on the issue. He explained that the biggest problem with first-generation (my term, not his) analytics like Corsi is that they treat all events as equal (a hit in the neutral zone, a blocked shot, a scoring chance on a breakaway and so on).
Now, along the lines of Campbell’s argument, Corsi is being worked over. Weighted shots are creeping into new fancy stats to value quality scoring chances more than wristers from the blue line. These fancy stat face-lifts should create a clearer picture of which players really drive possession and goal-creation.
Data Mining is Harder To Do Than You Think—and It’s The Future, Too
One theme that ran from start to finish was the painstaking, meticulous, mind-boggling work that analytics people do to gather their data.
Johnson noted that the data he compiles for hockeyanalysis.com comes from parsing a variety of sources, including Real-Time Scoring System (RTSS) information from every game, for every team, for every season.
Thomas discussed his work with War-On-Ice to rebuild a CapGeek styled website. He and his group are searching for primary data sources to build a replacement for the beloved/defunct CapGeek from the ground up.
And Emmanuel Perry of Senstats.com wowed/shocked the crowed by detailing his work analyzing “blue line events.” Perry is currently (and manually) tracking the 400-500 times that pucks cross the blue line in Ottawa Senators games to create an accurate view of zone time.
Perry appealed to the crowd for volunteers to help track games for the other 29 NHL teams.
The room filled with laughter. But, in a room filled with stats people, Perry may yet find some help. One way or the other, the average fan or blogger owes a lot of gratitude to the statisticians that tirelessly track every NHL event.
Fancy Stats for Goalies?
During the final panel, the veteran analytics stars (Vollman, Awad, Schuckers, and Seppa) handled a wide range of questions from the crowd.
The most interesting thread revolved around analytics for goalies.
Some adjusted save percentage metrics have surface (good examples can be found here and here).
But when a questioner asked the panel about how goalie analytics studies are coming along? The answer was…Bleak.
In paraphrase, Tom Awad answered that goalie data for saves facing a 2-on-1, stopping one-timers, handling screened shots, etc…are all pieces of data that just aren’t available.
Until that information is recorded and available, adjusted save percentages are the fanciest stats that fans can expect.
Though, as Rob Vollman concluded, the Fox Trax “glowing” puck was able to record some of the type of data that could be key for advanced goalies studies.
Leave it to a bunch of fancy stats nerds to argue for the return of the glow puck.
What do you think, analytics fans? Are you excited about the direction of advanced hockey statistics? Or, have fancy stats nerds become stuck on the minutia?