The following is an excerpt from the book "HockeyNomics: What the Stats Really Reveal," written by Darcy Norman. In this excerpt, Darcy takes a look into the history of hockey sabermetrics, why hockey and baseball are not comparable for the application of statistics, whether hockey is capable of catching up to baseball in the field of sabermetrics, and whether objective analysis can catch on in NHL front offices in the near future. For more information on the book, click here.
The quartet of [Alan] Ryder, [Iain] Fyffe, [Ken] Kryzwicki and [Chris] Boersma stand among a small group at the forefront of the statistical analysis of hockey. And although their work is largely unknown outside their inner circle, these men are slowly but surely continuing to dig through their data, shaping their spreadsheets and finessing their formulae with the hope of understanding hockey better than we do today. Their efforts are not unlike the humble beginnings of Henry Chadwick, many years ago.
Despite their varied backgrounds, these hockey analysts share some of the lessons learned from the insider-versus-outsider struggles first experienced in the baseball world. “This discipline is not meant to replace traditional methods,” explains Kryzwicki, “rather to enhance them. It can help corroborate or debunk a hunch or commonly held belief.” But even though they have confidence that their work is relevant, these men realize there is no point being militant about their methods because the old guard just won’t get it. “There are some people that if they don’t know, you can’t tell them. They’ll never be convinced, because they’ve decided that statistical analysis is useless, and no amount of evidence will convince them,”says Fyffe. “Those people aren’t the audience, and trying to make them the audience is futile.”
Part of their passive attitude exists because these analysts understand that baseball and hockey are completely different games, and the dynamic nature of hockey makes it harder to put numbers on the game. This difference has put hockey well behind baseball in terms of statistical understanding. “Baseball analysis has led the way because it is a game of a limited number of states,” Ryder explains. “It can be modeled accurately in discrete steps. This is not the case with hockey. Hockey is fluid and can only be modeled approximately.” Boersma agrees, stating that “Hockey might be the hardest sport to analyze. The parts of the game that are critical to a team’s success—like battles along the boards and puck movement—cannot be measured.”
No one can say, however, that the preliminary efforts of today’s hockey analysts have been in vain. It’s clear to these armchair analysts that what we currently have in terms of metrics isn’t going to cut it and can easily be improved upon. “Hockey is awash with meaningless and, even worse, misleading statistics,” Ryder sighs. He presents the goaltender’s win statistic as an example. “Someone decided to borrow this from baseball. But a typical team leans heavily on one goaltender, which makes the results of the team and goalie indistinguishable.” Rather than simply complaining about what we have to work with, Ryder has replaced the bad with the good, developing a much better measure of netminder performance: Shot Quality Neutral Save Percentage.
The NHL has not helped matters, though. The ease with which baseball can be described by numbers alone, combined with the insatiable appetite that its fans have for statistics, has forced Major League Baseball to be on the cutting edge of data tracking. Not so for data-hungry hockey fans. In fact, many analysts don’t see hockey analysis progressing much further if the NHL doesn’t pick up the slack. “If the data quality and elements get better, [hockey analysis] could become more mainstream,” Kryzwicki explains. He knows as well as anyone the limitations of the current catalog of statistics offered by professional hockey. Kryzwicki has spent years trying to develop a good draft model, even getting some interest from a scouting service, but he has never been able to complete his work. He knew what he wanted and how to do it, but “all I was missing was the proper data.”
Boersma considers this the main problem facing hockey analysts. Where does he see the field in 10 years? “Not much further. While each individual analyst will get better at what they do, they’ll still be limited by the information in the scoring reports. Only the NHL has the ability to increase the data.”
Fyffe isn’t so cynical. Recently drawn from an analytical slumber to write for the newly launched Puck Prospectus website, he sees the progress of hockey analysis as slow but typical. Fyffe points out how long it took rigorous baseball analysis to get a toehold despite the fact that “baseball has mountains of statistical information.” He says the reception of Puck Prospectus has been very good and believes that the NHL is moving in the right direction with respect to their data collection and distribution. “Before
1967, what do we have in a skater’s record? Games played, goals, assists and penalty minutes. That’s not much to work with. The introduction of real-time stats, allowing actual ice-time numbers, among other things, was a big step.” He holds out hope for headway within the decade.
Despite the complaints and annoyances, this small sample of hockey analysts agrees that they’re going to maintain the course. “A long time ago, I learned that if I’m doing something, I should first make sure I enjoy what I am doing,” says Boersma. And playing with hockey data is exactly such a thing. Boersma spends his spare minutes writing for his weblog, Hockey Numbers, complete with a database filled with unique statistics. “There are very few places you can find interesting statistical projects that can be analyzed every year and that produce such dramatically different results. Hockey’s also fun to watch.”
Fyffe is going to continue to dig into the data under a variety of guises. Foremost is the website, for which he was recruited, that is probably one of the most likely vehicles to bust this field into the mainstream. But he won’t neglect the Hockey Analysis Group, known as HAG, an old-school Yahoo! newsgroup he founded years ago that now boasts hundreds of members. “I don’t harbor any illusions that my work will revolutionize the way NHL teams look at their players or anything like that. I just know there are quite a sizable number of people who are interested in this stuff, so I like to share it with them.”
Darcy Norman is the author of the book "HockeyNomics: What the Stats Really Reveal."