June 10, 2015


The secret life and death of a hockey analytics pioneer (Marin Cogan, 5/29/15, ESPN)

FOR YEARS, THE hockey analytics community has lived almost entirely online, communicating via email or their blogs or a Yahoo group. Until the summer of fancy stats, many writers who had talked regularly for years didn't know anything about one another outside their work. Until they learned of Purdy's death, most of his fellow bloggers never even knew his name.

In this world, Tore Purdy was known exclusively as JLikens, the brilliant author of the blog Objective NHL.

Years after the Moneyball revolution, the NHL lagged miserably behind other sports in adopting analytics, and there were only a handful of writers -- Tyler Dellow, Gabriel Desjardins, Tim Barnes (writing as Vic Ferrari) and JLikens among them -- doing groundbreaking work. JLikens first came onto the blogging scene in 2008, launching Objective NHL with an article that briefly and elegantly disproved the commonly cited correlation between NHL goalies' save percentages and shots against. "There is absolutely no evidence that high shot totals have an inflationary effect on goaltender save percentage. Why, then, is it often argued that such a relationship exist[s]?" he wrote in the post. "More than anything else, the phenomenon seems to be driven by wishful thinking on the part of the claimants, a disproportionate number of whom belong to certain fan bases. I'll say no more."

From there, he quickly built a body of work that blazed a trail for the practitioners of advanced stats who came after him. "It's people like JLikens who laid the foundation. If there's a Mount Rushmore of hockey analytics, he'd be on it for sure," Yost says.

JLikens' grasp on mathematical modeling was strong, which enabled him to tackle the big questions -- questions that would seem merely rhetorical to the average right-brained reader, like "how often does the best team win?" He would pose a question that seemed unanswerable and then meticulously set about deriving a formula, running the numbers and coming up with the charts and tables he needed to prove it. JLikens presented his findings in the dispassionate, clinical tone of a researcher: "It turns out that the best team wins the cup 22 percent of the time -- about once every five seasons," he wrote, before outlining in detail the limitations of his research.

"He was certainly one of the first people to really get into doing what I call the hard proofs," says Dellow, another early analytics blogger, who now works for the Edmonton Oilers. "A lot of us would have ideas and kick them around. He came at it from a harder math angle, trying to prove some of the stuff."

A single JLikens article could articulate a subject-level mastery many people would struggle to convey over the course of their lifetime, and his blog felt like essential reading to anyone interested in advanced metrics. His breakthroughs helped spearhead the idea of studying statistics in games when the score was close, because leading teams tend to play more passively. He also advanced the idea of scorekeepers' home-rink bias and showed how significantly a team's stats can change depending on whether it's chasing or protecting a late lead.

"He basically verified everything we were doing in a rigorous statistical manner, in a way that made you comfortable with the results," says Benjamin Wendorf, a writer for The Hockey News. Concepts that hockey analysts now use regularly -- puck possession and how it's measured, shooting percentage regressions, shot quality -- "are only spoken about so confidently because a lot of his work has held up."

Most of JLikens' commenters were enthusiastic about his work, but when someone was uncivil, he wouldn't take the bait. Instead, he'd patiently dismantle the other guy's argument, burying him under an avalanche of facts. The only time he expressed impatience was when someone said something vague or euphemistic, like "good teams make their own luck." "What does that even mean?" he'd shoot back, before returning to his preferred mode of fact-based argument.

"A lot of people, when they're geniuses, like to use that to tell everyone else why they're wrong and flawed," says Rob Vollman, who runs the analytics site Hockey Abstract and contributes to ESPN. "He was purely seeking out insights. He'd never get distracted with what someone else was doing wrong." The passion in JLikens' work came through in its complexity, in the hours he'd clearly spent painstakingly running models and putting his posts together. "He'd be trying to answer a question with numbers and then just along the way he'd invent or create or discover or innovate something amazing, almost off-handedly, as an aside," Vollman says. "He'd create something that someone else would have to devote a lot of time to target. It was really a genius that he had."

Posted by at June 10, 2015 2:45 PM

blog comments powered by Disqus