The Sault Ste. Marie Greyhounds and their not-so-secret use of analytics
One of the first things Kyle Dubas did when he became the general manager of the Sault Ste. Marie Greyhounds was buy everybody in the office a copy of Jonah Keri's The Extra 2%, a book that chronicles the overhaul by two Wall Street analysts of the Tampa Bay Rays, turning them from a laughingstock of baseball to a perennial contender in the American League East Division.
It's not entirely out of place. The Greyhounds, competing in the competitive Western Conference of the Ontario Hockey League, face challenges as a smaller junior hockey market competing against money franchises like London, Windsor and Kitchener, teams that have a disproportionate amount of OHL success over the past decade, winning seven of the past 10 J. Ross Robertson Cup championships, and no other OHL champion has hoisted the Memorial Cup since 1990. (Though Ottawa and the Soo have both won as hosts in that time frame.)
Dubas, whose baseball geekery is well-documented on his Twitter account (he stays up past 1 a.m. some nights to watch the Seattle Mariners lose more often than win) gets that he needs to do things a little bit differently to compete. He greets competitors like the Knights as a challenge to overcome rather than a team to criticize for their aggressive approach to acquiring top talent.
He appeared on TSN 1050 in Toronto Tuesday afternoon with Gareth Wheeler and Jamie McLennan and spent a good portion of the segment discussing the future of the Greyhounds in light of his recent contract extension and how smaller market teams in the OHL can compete. Most of the discussion involved creating a team that players really wanted to play for, one that stressed academics and off-ice strength and conditioning programs. But he also suggested that in the end, players want to play for teams that win a lot of games and gain reputations for sending players to the NHL.
So the discussion turned real interesting when he was asked about his outspoken views on modern statistical hockey analysis. That begins at about the eight-minute mark of the link above:
It's a weird time in the hockey world Kyle, because there's a "great debate" I guess increasing… all to do with "advanced stats" and all to do with the way they're used in the game of hockey. I know following you you're a big advanced stats guy, yet to me that must be entirely challenging using advanced stats at the junior hockey level because it's gotta be hit or miss on the way that the stats are conducted or accumulated, so how do you balance using advanced stats as part of building a program versus just obviously your eyes or the natural feel for the game of hockey?
Right, well I look at advanced… analytics as just another avenue to broaden the way that we look at the game in a way that maybe we can find some inefficiencies with the way that teams evaluate players. I don't look at it as the be-all and end-all, I don't look at it as being right 100% of the time, and in our level what we have to do is we have to collect all the information and the statistics ourselves.
So shot attempts, zone entries, scoring chances, our league doesn't tabulate those like they do in the National Hockey League. So that falls on Tyson Enfield [Greyhounds manager of ticket sales & advanced analytics] in our office, and we usually have one intern per semester that will help us out with that. So what we're trying to do is we're trying to build a team that always has the puck, and I think that's what every hockey team in the end is trying to do and that's what I think the most… the analytics categories that have proven to be the most accurate are measuring, right? They're measuring which players have the puck the most when they're on the ice and which teams have the puck the most and throughout a season and over a given season.
And that's all we're truly trying to do is continue to advance that and try to find a way for us to improve the way we look at the game and the way that we look at players on our own team. We do a lot of self-evaluation on it because it's tough for us to calculate all the other 19 teams in our league, but I'm very… I have been for a long time looking at what people call "advanced stats" or "analytics" that are provided to people like Behind The Net dot ca or Hockey Analysis and a number of great people on the Internet but for me it's just another way to evaluate players and enhance our analysis of the game and for some people it might be overwhelming a little bit. The basic eye test will show for me that the best teams have the puck more, they create more offence, they're in their own zone far less than other teams and therefore when you're in that position you're more likely to score so if you can find a way to measure that it would be fantastic.
Ever since Moneyball came out in 2003, professional teams in all sports have been looking for that next edge, to compete like the Oakland Athletics did on a shoestring budget by finding undervalued players by looking at statistics like on-base percentage. In the last decade, "Big Data" has become somewhat of a buzzword and there are a large number of proprietary systems in hockey that some NHL clubs are using.
The Pittsburgh Penguins and the Vancouver Canucks are known for using certain systems. They're open about their use of research and analytics in making player decisions even if the methods themselves aren't open to the public. What's interesting about Dubas and the Greyhounds is that their method of statistical analysis operates in the public sphere—Kyle will frequently comment on stories dealing with analytics not just in hockey, but in basketball and in baseball as well. He follows a lot of writers of all sports, and has participated in Buzzing The Net's weekly chats in the past.
It's definitely in stark contrast to the way any other team in hockey works. Many online hockey analysts aren't career-focused or looking for employment with teams. They're hobbyists, writing because they're interested in the sport at websites like MC79Hockey (Edmonton Oilers-focused), Shutdown Line (Carolina Hurricanes), Broad Street Hockey (Philadelphia Flyers) and Fear The Fin (San Jose Sharks).
So perhaps it's surprising that Dubas is so open about his methods—they have to be the only CHL team that employs a manager of advanced analytics on its staff directory page—but like Dubas said, it's not about finding the "be-all and end-all", it's about a little extra edge. The public discussion in which Dubas and head coach Sheldon Keefe occasionally partake is one that has linked overall team shot attempts to offensive zone time, quantified the difference between a controlled zone entry and a dump-in, and found a correlation between a team's puck possession ability and its ability to create scoring chances.
Perhaps sometimes the information from the team itself can be a little overwhelming...
PRE 46.1%. POST 56.7% FenCl RT @camcharron: That @OHLHoundPower presser doesn’t have the team’s Corsi or Fenwick numbers pre-& post change.
— Soo Greyhounds (@OHLHoundPower) May 31, 2013
Jared McCann, named to the U18 Canada Camp today, had a Corsi of +29 over final 9 games. Also had 7g 9a playing top competition in that time — Soo Greyhounds (@OHLHoundPower) July 5, 2013
...but it shouldn't be suggested that the Greyhounds are "all in" on analytics since there are a lot of components involved with building a hockey team. Having as much usable information as possible about his own team can't hurt. It's just like the Rays, who found a way to improve everywhere they could, and wound up with a competitive, marketable team despite competing in baseball's richest division.