By Sunil Agnihotri
The field of hockey analytics took a big step with the recent release of Hockey Abstract, a book which aims to provide a guide to statistical analysis in hockey. As more and more people, including fans and professional teams, seek a deeper understanding of the game, hockey analytics continues to grow and develop.
Author Rob Vollman currently provides analysis for ESPN, Hockey Prospectus, the Nation Network and Arctic Ice Hockey. He was kind enough to provide some additional insight into hockey analytics and what its role is in the game.
Could you tell us how you got into hockey and hockey analytics.
Like many Canadians I’ve been playing hockey my entire life. Even in my adult years I’ve probably played well over a thousand beer league games. My love for sports analytics, on the other hand, began with baseball in the mid-1980s. It must have been strange for my parents to see their child immersed in numbers instead of video games and comic books.
I made the switch to hockey shortly after we baseball fans were betrayed in 1994. To this day I still can’t believe that what finally deprived us of a World Series wasn’t a war, an economic depression, an epidemic, a gambling scandal, or even a natural disaster, but rather a labour dispute between millionaires.
As for hockey, I found Iain Fyffe in the online world about 13 years ago. My first published hockey analytics article was in the Hockey Research Journal in 2001 a collaboration on fixing the plus/minus statistic. Isn’t that always the topic hockey analytics folks choose first? I worked with Iain, as well as guys like Tom Awad, Alan Ryder and Gabriel Desjardins, especially after Iain launched a Yahoo discussion group in 2004. When Hockey Prospectus launched in 2009 its founding editor Andrew Rothstein brought us all in (except Alan), and that’s probably when my work started finding its voice.
This book was clearly a lot of work and brings together numerous ideas regarding hockey analytics. What was your intention publishing the book? What motivated you personally to work on a big project like this?
Bill James and his Baseball Abstract meant a lot to me when I was getting started, and I always felt the absence of something like that in hockey. There was Klein and Reif’s Hockey Compendium around that same time (1986), but really nothing else. Ten years ago Iain, Tom and I actually discussed plans of writing a book like this, and over the years several other famed analysts have also tried to get a project like this off the ground, but the interest hasn’t really been there until quite recently.
There was no particular inspirational event that finally triggered it, just that growing desire to have a book like this out there – combined with my own growing number of things to say. I guess you could say that everything finally reached a boiling point. Young hockey fans today need the same kind of books we had for baseball thirty years ago. And more mature fans needs the guide to help them understand and apply this aspect of the sport … and win their fantasy hockey pools, of course.
We’re seeing hockey analytics get used by teams and fans for numerous reasons. Why do you think hockey analytics is important?
Good question. Hockey analytics can be used as a sober second thought to validate what we see with our eyes, and in a way that’s hopefully free of emotion and bias. Sometimes they can also uncover things that we’ve overlooked with more traditional analysis, and that are deserving of some closer attention.
Ultimately no one can watch all 1,230 regular season NHL games, plus play-offs, not to mention the AHL, Canadian Juniors, U.S. College, and all the European leagues. The value in finding ways to mine all that information for whatever interests us most is so obvious that perhaps the real mystery is why it wasn’t embraced sooner.
Why do you think hockey analytics is growing? And, In your opinion, what does the growth of hockey analytics say about the game and its fans?
I believe the growing demand for hockey analytics is part of a greater movement. For example, we saw Nate Silver of the New York Times famously use analytics to successfully predict the outcome of the last U.S. election – and now he’s with ESPN.
I think hockey fans have seen analytics in so many different places that they’ve not only grown a lot more comfortable with it being applied to hockey, but have even become eager to see what they can contribute. In one sense the pressure it really on right now. We’ve got their attention, what are we going to say?
In your opinion, what factors are influencing the growth of hockey analytics?
Access to information has been a big driving force in the growth of hockey analytics. While it was quite difficult for people like Iain, Tom and me to find each other and the data that we wanted ten years ago, now anyone can get their hands on most statistics relatively easily. There’s a critical mass that’s being reached where the analytically-inclined can find each other, and build on each other’s work. I can only expect that to keep growing, and I’m always excited to read about what’s coming next. (Hopefully not something about how the Leafs are clutch).
You continue to collaborate with the online community of fans, who, in my opinion, are really pushing the hockey analytics field.
Yes, I’d agree that it really seems to be the fans that are driving the growth of hockey analytics right now. It’s great to have access to such a large community of fans who will look at what you’re doing and tell you what works, what doesn’t work, and how to make it more useful and relevant.
How much has the online community influenced your work and your understanding of the game?
The most significant influence has been on focusing the direction of my work, helping me decide which of my studies had the most merit. I think I also learned more about how to strike the right tone when talking about a sensitive topic like hockey analytics. My work is often best presented with a voice of curiosity, and sometimes I take great pains to eliminate even a trace of something that can be perceived as know-it-all condescension. It’s not easy, and I don’t always pull it off, but I have the most success reaching fans when I do. At least those that aren’t complete morons.
Your visualization of hockey data has been a remarkable tool to help simplify the data. Providing an easy-to-use tool to generate custom player usage charts is also a great way to teach people about the game. To me, this can help reach out to more people in the game and make hockey analytics a little more mainstream.
I was actually a little surprised just how much it helped reach more people by translating our tables of statistics into charts and graphs, like the Player Usage Chart. I was also very lucky to meet a visualization expert like Robb Tufts who can help bring so much of our work to life (like, most recently, the team luck chart).
I’ll definitely be integrating more charts and graphs into future work. I’ve already introduced Special Teams Usage Charts which show a player’s average power play time on one axis, penalty-killing time on the other, and a bubble sized by their even-strength playing time. It really drives home the message of the kind of ice time a player is getting. I also brought out a Total Offense Chart which shows shots per 60 minutes on one axis and passes per 60 minutes on the other, which capture the picture of who is driving a team’s offense surprisingly well.
Is it important for hockey analytics to become more mainstream?
To answer your question, let me first bring up a philosophical discussion that’s currently taking place within the analytics community. One side believes that our studies need to be made complete and foolproof before they’re introduced outside our community. The argument is that flawed studies will have results that may discredit everybody’s work. On the other hand, a complete and complex new development may require too much effort for a so-called mainstream fan to understand and embrace – and the allegedly foolproof nature may come across as an arrogant turn-off.
The solution is probably a balanced approach. Introduce new analytic work even in its earlier, simpler and more incomplete stage, but be very honest and transparent about its applications and limitations. Oh, and we should rename Corsi (just kidding!).
Do you foresee any barriers that might slow down hockey analytics today or in the future?
No. Even poorly-named statistics, arrogant analysts, old-school stigma and flawed studies can’t de-rail the growth of hockey analytics. There’s just too much potential to slow down the momentum. Or at least that’s what I tell my publicist.
Your book is a pretty big step for the hockey analytics field as a whole. What do you think could help push hockey analytics to that next level? Some have suggested better technology to capture real-time data (i.e., ice-tracking).
Reaching the next level will take time, and for everybody to buy my book. Actually in all seriousness, there are a lot of brilliant minds with a lot to contribute to the field of hockey analytics. Even if it isn’t perfect, the success of one book might help bring us more – instead of this just being a one-time Klein and Reif Hockey Compendium all over again.
But yes, we also need to do a more accurate and consistent job in gathering the data that we already do, adopt the same approach with other leagues, and start finding new valuable types of data to collect. Perhaps it will actually be the fans doing most of this.
Thanks very much for your insight, Rob. Your book is getting a lot of attention right now for good reason. We wish you continued success in the future!
Thank you for the kind words – and thank you for this interview, I appreciate your encouragement and support.
Fischer, J. (2013, July 31). Book Review: Rob Vollman’s Hockey Abstract. In Lou We Trust. Retrieved from http://www.inlouwetrust.com/2013/7/31/4573480/book-review-rob-vollmans-hockey-abstract
Horner, M. (2013, July 31). Book Review: Rob Vollman’s Hockey Abstract. Five Minutes for Fighting. Retrieved from http://www.fiveminutesforfighting.com/2013/07/book-review-rob-vollmans-hockey-abstract.html