April 2012 Past Issues | Advertise | www.cia-ica.ca
Return to main page

Hockey-Loving Actuary Scores With Statistics

Print Print this Article | Send to Colleague



The nation is once again suffering Stanley Cup fever, with the play-offs putting countless hockey fans on the edge of their seats.

Like millions of others, actuary Alan Ryder will be analyzing the games and hoping to predict the outcome. However, unlike those millions, he will be using statistical and actuarial expertise that led to him being approached by NHL teams eager to benefit from his knowledge.

The CEO of Aurigen Reinsurance, Mr. Ryder, FCIA, FSA, MAAA, is renowned for his website hockeyanalytics.com, which attracts hockey lovers keen to discover the numbers behind the scorelines.


Thanks to his background in statistics and a career as an actuary, Mr. Ryder has been able to produce detailed calculations that can reveal the most important players on a team and much more. His site covers such topics as the applications of the Poisson distribution, win probabilities, the "five laws of hockeynomics", and the impact of puck possession.


All this is a long way from 2002, when he was inspired by the book Win Shares. Written by Bill James, the grandfather of baseball analytics, it set out to allocate team wins to individual players. A lifelong hockey fan who enjoyed recording such data as line combinations and the location of shots on goal, Mr. Ryder devised Player Contribution, his own system for determining the contribution of individuals to a team’s wins.


His model of shot quality was groundbreaking work. About it he said: "A simple model for the allocation of goal prevention accountability is that ‘defense prevents shots’ whereas ‘goaltending stops shots’. But all shots are not the same. There are significant variations in quality. So I set out to improve on this simple model by building a predictive model for goals based on observable shot quality factors such as distance and shot type. This shot quality model assessed the factors and produced the probability of a goal.


"As a good actuary I wanted the model to be ‘reproducing’—the model’s expected goals should equal the actual goals while even-handed, on the power play, and while short-handed, and across all shot types. When you sum up the shot probabilities for each shot in a game you get the expected goals in that game. This thinking ought to be familiar to any actuary. And this thinking is now used by several NHL teams to better assess team and individual performance, as the number of goals actually scored has a significant random element."


Hockeyanalytics.com has become so well regarded that Mr. Ryder has been offered consulting assignments by NHL teams, and has enjoyed one-on-one sessions with some of their senior managers. However, he prefers to treat the site as "an enjoyable hobby".


Years of analysis have meant that he now views games differently. "When I was younger I used to follow the offense. But now I watch the defence a lot more. Defence is the least understood part of the game because there are no readily available statistics that describe it. That’s not the case with forwards—conventional statistics help us there. How does a fan know whether a defenseman is playing well? They really don’t have any guidance.


"Actuarial science has given me a very specific framework to think about problems. One of the things that you need to study hockey through an analytic lens is an understanding of randomness. The puck bounces around a lot. Sometimes it finds its way into the net. Without an understanding of probability, you may not know what the numbers mean."


The baseball-centred book Moneyball, which was the basis of a hit film starring Brad Pitt, has brought the area of sport-related analytics to wider public attention. Mr. Ryder said: "It popularized the notion that you could get somewhere with this thinking. And as a business book Moneyball is excellent, as it has a clear message: there is something to be gained in the identification and acquisition of undervalued assets.


"Hockey is a difficult game to measure. You could easily model baseball on a computer, as it has so many discrete events. But hockey is the most fluid team sport. And so much of it is a function of energy and passion."


The passions surrounding hockey have brought Mr. Ryder into conflict with some fellow fans. During the three seasons he spent writing an analytical column for GlobeSports.com, which was open to readers’ comments, he had to develop a "very thick" skin. "Many fans do not believe that statistics can be useful, and most people want to believe their eyes, not the numbers. But, for instance, the difference between average and marginal goaltending might be only one goal every three or four games. You have to watch a lot of hockey to see that with your eyes. There is, however, a thoughtful community out there of other hockey analysts, sports writers, bloggers, and teams. Those people are really interested."


Like every other supporter, Mr. Ryder has an opinion on this year’s Stanley Cup. Describing himself as "a Leafs fan, unfortunately" ("I believe you should pick a team and keep supporting it"), he said: "At the end of the regular season I thought the best team was the Pittsburgh Penguins, but they are gone. Hockey is full of random events and the best team does not always win a period, a game or a series. I think St Louis could really surprise people. They are exceptionally disciplined and they win games with defence. They keep games under their control. Their goaltending is hot, which can carry you through the play-offs."


However, he added a note of caution: "There are so many ways it can go wrong."

 

360 Albert Street, Suite 1740, Ottawa, ON K1R 7X7
Tel.: 613.236.8196 Fax: 613.233.4552
www.cia-ica.ca