While there are many short books to be written about hockey — Breaking Down Every Shift of Scott Cullen’s Pro Hockey Career or An Anthology of Toronto Maple Leafs Playoff Games Between Lockouts — there may be a new contender: The Hockey Panel at the 2013 Sloan Sports Analytics Conference.
Okay, that may be tad dramatic but, after what seemed to be successful panels in the past two years (Hockey recap 2011 and 2012 Recap), there was no hockey panel at this year’s conference, bringing disappointment to the stick-and-puck crowd.
The only panel I attended that included former Maple Leafs GM Brian Burke, Break-Ups in Sports, wasn’t very hockey-oriented and Burke re-iterated some of his tried-and-true anti-stats soundbytes from the year before.
Considering that the hockey panels had been well-attended the last two years, it was disheartening, but there were still some fascinating hockey subjects discussed. (You can see, from this year’s overall conference review, that I felt there were valuable discussions anyway.)
For example, Eric Tulsky presented a research poster (worked on by Eric with Geoffrey Detweiler, Robert Spencer and Corey Sznajder) on zone entries that revealed how valuable it is to enter the offensive zone in control of the puck, as opposed to employing a dump-and-chase technique.
Certainly, there are times (eg. outnumbered by defenders, line change etc.) when dumping the puck into the offensive zone may be the most reasonable play but, circumstances being equal, teams were about twice as likely to generate a shot on goal when they carried the puck into the zone and that didn’t vary much based on the skill of the players involved. As Eric wrote about previously, Jaromir Jagr and Zac Rinaldo generated shots at the same rate for the Philadelphia Flyers last season when they entered the offensive zone with possession. Since there is a rather wide talent gap between the two, it then becomes relevant to note that the frequency with which they entered the offensive zone in possession of the puck was vastly different.
While some advanced stats might appear counter-intuitive, the zone entries paper totally fits with expectations of the logical hockey observer. The odds of keeping possession of the puck, when you already have it, is of course higher than it would be to retrieve a dumped-in puck that, effectively, becomes up-for-grabs.
Tracking these movements requires a signficant commitment, so kudos to those who did the work and while I joked with Eric that I tried to track zone entries while watching the Bruins and Senators last Thursday, this information can change the viewing experience.
When you become increasingly conscious of who is driving the play into the offensive zone, and holding possession of the puck, it enhances appreciation for the players that do it well (in that particular Bruins game, Patrice Bergeron was notably strong in this regard).
I know the guys at the Cult of Hockey, in Edmonton, have been tracking zone exits for the Oilers and, again, this seems to be useful information.
Getting this data on a league-wide basis, of course, would be fantastic.
TOTAL HOCKEY RATING (THoR)
The one other official conference hockey contribution was Michael Schuckers’ Total Hockey Rating (THoR), presented by Michael Schuckers, an Associate Professor of Statistics at St. Lawrence University who is the Co-Founder of Statistical Sports Consulting and presented a Defense Independent Goaltender Rating at the conference a couple of years ago.
The premise behind the Total Hockey Rating was to assign a value for every stat on the NHL’s Real-Time feed, and its location on the ice, to identify its correlation with goals being scored.
In theory, it seems like a sound approach and when someone respected in the stats community like Tom Tango weighs in with some cautious optimism about the process, that’s all the more reason to consider it.
That said, THoR came under immediate fire online because it revealed that the third-best 5-on-5 player over the 2010-2011 and 2011-2012 seasons was…Pittsburgh’s Tyler Kennedy, a fine third-line winger who has been generally unremarkable to the general public in his five-plus years in the NHL.
By no means would I rank Kennedy in that spot, but I can see some of the forces that were at play.
For one thing, the process, removing the variability of shooting percentage, focuses on shots (and locations) and that helps Kennedy, who scored on a meagre 7.5% of his shots over those two seasons, ranking 227th among 230 forwards (Brian Rolston, Darren Helm and Nathan Gerbe were the only ones worse) that scored at least 20 goals over those two years.
So, by removing shooting percentage (or goals) from the equation, a shot from Tyler Kennedy is considered the same as a shot from Steven Stamkos, even though Stamkos had scored on 18.3% of his shots through those two seasons.
(By the way, I recognize that I’m using overall shooting percentage as opposed to 5-on-5, but the point essentially, much like the song, remains the same.)
This is particularly important in Kennedy’s case because he generated a lot of shots. He ranked 28th in the league in shots on goal per game over those two seasons (3.06 per game) and that’s impressive enough, but he also did so in a fraction of the time that other leaders played.
Over those two seasons, Kennedy played 2025 minutes. Of the 58 players to record at least 400 shots on goal, though, only one — Evgeni Malkin — recorded more on a per-60-minutes basis and Kennedy didn’t have the benefit of significant power play time to increase his offensive numbers.
Here are the leaders in shots/60 minutes for forwards with at least 400 shots on goal in 2010-2011 and 2011-2012 combined:
Kennedy sticks out like sore thumb in this group, too, so I see how his ability to generate shots on goal could have sent his ranking to a level that no one would have anticipated.
That doesn’t mean, however, that the idea behind the rating system isn’t a good one. Refining, and including power play and penalty-killing values, could eventually present a list that seems more representative of player values and, at that point, can be utilized to help find assets that may be undervalued. I’m a sucker for player rankings, having provided them on TSN.ca for more than a decade, so I will readily keep tabs on further developments in this field.
Of course, my frustration from the get-go with the Sloan Conference is that NHL teams don’t seem to be as accepting of analytics as I am. I’m numbers-oriented, pretty much always have been, so I get that I’m more open to some of the analysis presented, but I’m always hoping for some sign that NHL teams are taking analytics as seriously as teams in MLB and NBA, two sports that seem to be most accepting.
Even if Gabe Desjardins, of Behind the Net fame, says that the number of teams hiring a blogger to provide consulting services is growing, that pales in comparison to MLB and NBA teams, at the very least, in which most not only have someone working full-time on analytics but, in more and more cases, have actual analytics teams, so that they might have half a dozen people working on numbers, salaries and player evaluation, trying to find an edge.
Eric Tulsky went through NHL team pages and found four teams that employ someone in an official analytics capacity. There are certainly others, with job titles that don’t necessarily include analytics, but it’s not like it’s the industry standard for an NHL team to have a full-time analystics staff. By my count, there were five teams (Dallas, Edmonton, Tampa Bay, Vancouver and Washington) represented at this year’s conference by someone in hockey operations, so I’m left feeling like there’s a lot of room for growth, both in the conference’s hockey content and in NHL teams’ willingness to consider analytics as part of their everyday operations.
I don’t know if the cost makes sense to have six or seven analysts, but it remains unfathomable to me that teams don’t, as standard business practice, pay $100,000 or $200,000 a year to a couple of people who can help them make better decisions on payrolls that are between $50-million and $70-million a year.
While there weren’t any official hockey panels convened at Sloan, there were several informal gatherings both Friday and Saturday night. Schuckers and Brian Macdonald have been trying to push hockey analytics at the conference in the three years I’ve attended, so I enjoy discussing their analysis. It was good to meet Tulsky, a chemist in the real world, who sees how much room for improvement there is when it comes to the use of analytics in the NHL.
NHL statheads can dare to dream, right?
Scott Cullen can be reached at Scott.Cullen@bellmedia.ca and followed on Twitter at http://twitter.com/tsnscottcullen. For more, check out TSN Fantasy on Facebook.