Hockey Confidential Read online

  CHAPTER 4

  #fancystats

  Colliding Worlds and the Surprising Real Story of Corsi,

  Fenwick and PDO

  * * *

  All these years later, I owe my high school math teacher an apology, because I’m about to break a long-standing promise.

  Sorry about that, Mrs. Uyenaka.

  It was the final week of my Grade 12 year, June 1974, at Woburn Collegiate Institute in Scarborough, Ontario. As I recall, I was in a state of high anxiety because the list of students who would be obliged to take the year-end final exam was soon to be posted. In order to be exempt from taking the exam, a student required a mark of at least 55 per cent.

  It was going to be touch and go for me.

  Math—or science, for that matter—had never been my strong suit. The written or spoken word? Write an essay or make a speech? Bring it on. An equation or formula? Find a solution? Get out of here. My brain is not wired to deal with it.

  I approached Mrs. Uyenaka the day before the list was to be posted and asked her if I was going to make the cut at 55. She looked it up in her book, looked at me and shook her head.

  “No, I’m sorry,” she said. “You have 53 per cent. You have to write the exam.”

  I might not have known what a logarithm was—still don’t, actually—but I damn sure knew if I had to write that final exam, I was not going to pass it. If I didn’t pass the exam, I would fail Grade 12 math. If I didn’t pass Grade 12 math, I wouldn’t get the required credit for my high school diploma. If I didn’t get my diploma—well, if you knew my mother, that was not an option.

  I’d already chosen my six courses for Grade 13—two English, two history, French and family studies—so it wasn’t as if I actually needed Grade 12 math as a prerequisite for anything, other than getting my diploma and graduating in good standing.

  I showed Mrs. Uyenaka my option sheet for the next year, explained to her my innate inability to process numerical data, guaranteed her I could not pass the exam, told her there was no benefit to anyone—least of all me—to my writing that final exam, and then made her a solemn promise in the form of an offer I was fearful she could refuse.

  “If you bump up my mark to 55 and exempt me from the final exam,” I pleaded, “I’ll never, not ever, have anything to do with numbers or math for the rest of my life. I promise.”

  She said she would think about it. The next day, she posted the class list with our marks alongside our names: Bob McKenzie 55.

  Exempt.

  I’ve never forgotten what Mrs. Uyenaka did for me that day, nor have I forgotten or broken the promise I made to her back then.

  Until now.

  Damn you, #fancystats!

  We will likely look back one day on the 2013–14 NHL regular season as the proverbial turning point, the year in which advanced statistics in hockey—a.k.a. #fancystats or analytics (Corsi, Fenwick, PDO, etc.)—went mainstream. Maybe “mainstream” is a bit of a stretch, but there most definitely was an awakening. A line was crossed. Advanced hockey stats became more of a talking point, started showing up more often in more prominent places. The debate over their merits, or lack thereof, became louder and longer and more spirited, waged in newspapers and on television, radio and the Internet and social media.

  For that, we can thank the Toronto Maple Leafs.

  The Leafs, and many of their fans, believed their 2013 playoff appearance—the historic third-period-and-overtime Game 7 meltdown against the Boston Bruins notwithstanding, to say nothing of the team’s first playoff date since 2004 coming on the basis of a lockout-shortened 48-game regular season—was a portent of good things to come, a launching pad for a team headed in the right direction. But the purveyors of #fancystats said it before the 2013–14 season even began: the Leafs were cruisin’ for a bruisin’. Their “puck possession” numbers (as measured by tools such as Corsi and Fenwick) were way too low; their save percentage and shooting percentage (PDO) was unsustainably high. It was, the hockey eggheads maintained, a perfect statistical storm.

  The battle lines were clearly drawn. The Leaf season, for better or worse, was going to put #fancystats on trial.

  That this was playing out in Toronto, of all places, only raised the stakes. Next to Mayor Rob Ford (no comment), Corsi and the Leafs might have been the hot-button topic in Canada’s largest city in 2013–14.

  “I don’t think there’s any question about that, the [Maple Leaf angle] pulled [advanced statistics] into the spotlight,” said Tyler Dellow, a Toronto-based lawyer and blogger (his blog can be found at www.mc79hockey.com) who has emerged as one of the foremost authorities on the use of analytics in hockey. “The Leafs are a big deal and Toronto is the centre of the [hockey and hockey media] universe. They raised the profile of the debate to a level that couldn’t have happened anywhere else.”

  The mercurial Maple Leafs, meanwhile, cooperated by providing a dramatic, season-long script. They went 11–6 in their first 17 games. Maple Leaf general manager Dave Nonis, speaking at a sports business management conference at that point in the schedule, made a remark along the lines of having had teams in the past that outshot their opponents and lost, but now the Leafs’ Corsi—in his word—“sucks,” yet the team was winning.

  For many in the mainstream media who either didn’t like the creeping presence of #fancystats or perhaps just didn’t like their zealous proponents, it was open season on the newfangled numbers. As the season wore on, the Leafs certainly looked like a playoff team. With less than a month to go, they had 80 points in 68 games. They sat third in the Eastern Conference and ninth overall in the 30-team NHL.

  Critics of #fancystats were giddy with delight. They started doing their touchdown dance at the 10-yard line. We all know what happened next.

  Incredibly, the Leafs lost eight straight games in regulation time, and then briefly stopped the bleeding with a pair of wins, before closing out the regular season with another four straight losses in regulation. They lost 12 of their final 14, all in regulation time, falling from third in the Eastern Conference to 12th in the span of less than a month. The Leafs’ playoff hopes were incinerated into a mushroom cloud.

  It was one of the most epic collapses in NHL history. It was also taken as vindication and massive victory for the #fancystats gang.

  “As a person with some investment in seeing hockey analytics become more widely accepted, watching the Leafs collapse in slow motion after a season of taunting from the more traditional corners of the game was exceedingly gratifying,” Dellow said. “As a fan of irony, it might have been even better.”

  Dellow said the Leafs were exceedingly fortunate to have piled up 80 points in their first 68 games; still, it was reasonable to expect they would have played well enough in the remaining 14 games to make the playoffs. That they didn’t, Dellow said, was as much bad luck as anything else.

  “They died as they lived—on the bounces,” Dellow said. “While the numbers guys were vindicated, in that we’d correctly identified the Leafs as a team that wouldn’t make the playoffs, it also served as a reminder that you can’t say when the luck will run out, just that it will.”

  It mattered little that similar #fancystats forecasts of doom and gloom for the 2013–14 Colorado Avalanche never came to pass, that the Avalanche posted terrible Corsi, Fenwick and PDO numbers but dodged all the bullets and still finished with 112 points, second in the powerful Western Conference, third in the entire league. In advanced stats, as in any sport, you win some, you lose some. Some victories, though, end up having a greater impact than just another two points in the standings. Toronto’s collapse, the fulfilment of the Leafs’ #fancystats prophecy, was the analytics equivalent of a franchise-defining, we-walk-together-forever, last-second win for the ages.

  The truth about advanced stats in hockey is that they’re not really all that advanced. (This from the guy who had to beg his way out of a Grade 12 math exa
m).

  Advanced is more a relative term, given that goals, assists, points, penalty minutes and, more recently, plus-minus, have always been the standard currencies by which individual players were evaluated. Wins and losses, meanwhile, were always the norms by which teams are judged.

  But if advanced stats in hockey aren’t really all that advanced, they are most certainly new to the game—at least, relatively speaking.

  Many proponents of #fancystats point to the advent of Corsi, the puck-possession metric, as perhaps the dawn, or at least the big breakthrough, of modern-day hockey analytics. There was no specific date when Corsi was born, though Dellow recalled the concept might have come up on message board chatter (on sites like the HF Boards at hfboards.hockeysfuture.com) amongst the number-crunching community as early as the NHL lockout of 2004.

  Perhaps the first documented evidence of the actual statistic now known as Corsi—arguably the seminal moment in #fancystats history—appeared in 2006. Legend has it that Edmonton Oiler fan and blogger Vic Ferrari—the man, the myth, the legend— heard Buffalo Sabre goalie coach Jim Corsi explain in a radio interview how he counted shot attempts (totalling shots on goal, missed shots and blocked shots) instead of the conventional (and official) measurement of shots on goal. The story goes that Ferrari took that notion, worked with it and voilà: Corsi—the concept, not the man—was born. It’s believed Ferrari first wrote about it in any length on March 5, 2006, on his Irreverent Oiler Fans blog.

  Gabe Desjardins, a native Winnipegger, lifelong Jets fan and electrical engineer who graduated from Queen’s University in Kingston, Ontario, but moved to California’s Silicon Valley to work in high tech, believes the #fancystats movement had much deeper roots, starting in almost prehistoric hockey times. As in the 1950s. Desjardins would know: he is, for all intents and purposes, one of the godfathers of hockey analytics, the individual (along with Corsi creator Ferrari) who has done the most groundbreaking work to pioneer the cause. He started writing about hockey analytics in 2003. In 2006, as Ferrari was writing about Corsi, Desjardins established Behind the Net (www.behindthenet.ca), the hardcore hockey analytics website. There isn’t anyone today within the advanced stats community who doesn’t pay homage to Desjardins as a master.

  “We know the Montreal Canadiens of the 1950s were tracking plus-minus as a statistic long before it came into existence [in the NHL],” Desjardins said. “Some scorecards from the 1972 Summit Series have been found that showed [Team Canada coach] Harry Sinden was having someone [Ron Andrews, then the NHL’s statistician] count shots in a ‘Corsi way.’ We know Roger Neilson was scoring his players using shots for and shots against, so it’s been going on for a long time.”

  Incredibly, it appears at least one man, a true visionary, was looking at hockey in a complex, analytical, statistical way as far back as the early 1950s. Lloyd Percival—the man who pioneered everything from cutting-edge athletic training to injury treatment, to nutrition, to coaching methods, and the author of The Hockey Handbook, which Anatoly Tarasov, the godfather of Soviet hockey, credited as the blueprint for the development of Russian hockey—was doing work for the Detroit Red Wings in the early 1950s, but also for the St. Michael’s Majors Junior A hockey club. In Gary Mossman’s fine biography Lloyd Percival: Coach and Visionary, there is evidence that Percival was breaking down hockey games in a truly advanced analytical way. Not relatively advanced for the 1950s; advanced by even today’s standards.

  Of Percival’s work with St. Mike’s, Mossman wrote:

  Percival was able to do things for [St. Mike’s] that he was denied in Detroit. For example, he produced a seven-page “Hockey Survey Analysis” of a playoff game between St. Michael’s and St. Catharines on March 19, 1952, in which body checks for both teams are recorded, categorized, counted according to location on the ice and the level of success, and connected to scoring chances. Shoot-ins and recovery rates are also totaled, positional play is analyzed and passes are tabulated according to their locations, their success rate and the reasons for their success. Furthermore, shots at and on goal are counted and the location from which they were directed is analyzed, and the time of puck possession inside the blue line is recorded for each team. Along with conclusions in each section, Percival presented [St. Mike’s] with a summary of twelve “General” comments and recommendations for tactics and improved play for the next game in the series.

  Not only was Percival apparently embracing rudimentary Corsi in 1952, he was light years ahead of his time on zone entries and recovery rates, to say nothing of coming up with tangible tactical changes for the next game based on stats compiled from the previous game. Percival’s level of statistical sophistication was nothing short of incredible.

  In more modern times, Desjardins also cited the pioneering concepts and written work of people such as Tom Awad and Alan Ryder, from the 1990s through to the present, as ample evidence of thinking-outside-the-box hockey statistics that pre-date Corsi. Also, the NHL’s move to real-time stats around 1997 was a factor, as was Desjardins’s own involvement in an advanced sports stats/analytics site—Protrade, which later became Citizen Sports, formed in late 2003 and early 2004, where Desjardins was the lone hockey “expert” amongst others doing the same work for baseball, basketball and football.

  Still, Desjardins, like most everyone else involved in hockey analytics, points to the arrival of Corsi by way of Ferrari as a watershed moment.

  Whatever you think of advanced stats in hockey, the story of Ferrari and how Corsi came to be is a mindblower that most of the #fancystats community would not believe.

  Ferrari and some fellow “hockey nerds” (as he affectionately termed them) were discussing, in or around 2005, the concept of how one might efficiently measure puck possession, lamenting the absence of stats (blocked and missed shots) that would permit data to be collected. Then Ferrari heard Buffalo Sabre general manager Darcy Regier—not Jim Corsi, as legend has it—talk in a radio interview about missed and blocked shots, precisely the path Ferrari and his pals were interested in following.

  In a rare—if not unheard-of—telephone interview with me in April 2014, Ferrari explained publicly (for the first time) what happened after that:

  I heard Regier on the radio. He had all these great stats about missed and blocked shots that we just didn’t have at the time, although they became available [from the NHL] not too long after that. There was a small group of us that had been talking about [the concept] for a while.

  I was going to call [Corsi] the Regier number. But it didn’t sound good; it didn’t seem right. Then I was going to call it the Ruff number [after Sabres coach Lindy], but that obviously sounded bad. So I went to the Buffalo Sabre website and looked at a picture of a guy on their website, and Jim Corsi kind of fit the bill. So I called it a “Corsi number,” and then I pretended it was him I heard him on the radio talking about it—that’s what I told people. That’s basically [how Corsi got named].

  Wait a minute. Pump the brakes. This qualifies as breaking #fancy­stats news.

  Was Ferrari actually saying Jim Corsi’s only connection to the statistic was that Ferrari liked the look of his photo on the Sabres’ website, especially his moustache, and the sound of his name?

  From Ferrari’s perspective at the time, yes, that’s entirely accurate.

  “I always prepared myself [that if the stat ever became well known]—hey, it was just a small group of nerds talking hockey—that eventually Corsi or someone would come to me and say, ‘What the hell are you guys talking about and why are you [using my name]?’ I figured if it happened, I would apologize and carry on. I was really surprised a few years ago when I read a story in USA Today where Corsi talked about how the inspiration [for the numbers] came to him when he was skiing in the Alps, and I thought, ‘[Expletive deleted], it came to me when I saw your picture on a website, because I liked your moustache.’”

  But the story gets even better.
/>   Ferrari had no idea at the time—nor even when he gave this interview in 2014—that the metric’s name turned out to be fortuitously labelled. One of the reasons Regier was on the radio talking about shot attempts in the first place was that Corsi was, in fact, a believer in measuring a goalie’s workload not by just shots on goal but by blocked and missed shots as well.

  “Oh, I had no idea of that,” Ferrari said. “I just liked his moustache.”

  Seriously, you can’t make this stuff up.

  But it’s true, all of it.

  Regier was an NHL general manager who was in on the ground floor of analytics, and his goalie coach, Corsi, was a thoughtful, deep thinker, a former algebra teacher.

  “I always kidded Jim that he was the self-proclaimed protector of all goalies,” Regier said. “He was always looking for a stat that would give his goalies their due. [Adding up shots on goal, blocked shots and missed shots] was something along those lines. Jim was always charting shots—where they came from, that stuff. In all the years I’ve known him, Jim never tried to take credit for [the Corsi metric as it’s more sophisticatedly applied now]. He was just interested in tracking shots for his goalies.”

  Regier said there was other outside-the-box thinking going on within the Sabres organization at that time, mostly out of necessity.

  “Our scouting budget had been really slashed,” Regier said, “so we hired a bunch of young kids to track things [from games] on video. I can assure you, if I was on the radio talking about that sort of [statistical] stuff, it would have come from Jim and those kids and the work they were doing.”

  If the story of how Corsi came to be named is fascinating, what are we to make of the story of the mysterious, international man of mystery, Vic Ferrari?

  If you happened to be a fan of the 1980s TV sitcom Taxi, you’ll recognize the Vic Ferrari moniker. A character on the show, Latka Gravas—played by eccentric actor Andy Kaufman—created a suave but obnoxious alter ego who went by that name.