The worst thing you can possibly do if you’re going to be taken seriously in a discussion of hockey analytics is take an arbitrary 30-second sequence that became a critical turning point in a single game, and attempt to extrapolate larger truths from it.
That said, there are five minutes left in the first period on Saturday, Oct. 26, and the Pittsburgh Penguins are on the power play in the Air Canada Centre with a 1–0 lead.
The Penguins are dominating the puck. It leaves the Leafs’ zone twice on this man advantage, both times as the result of an errant pass back to the point getting past a stick. The Leafs do not clear the puck so much as breathe a ragged sigh of relief when the Penguins accidentally do it for them. The Pens cycle the puck flawlessly on the perimeter, taking their time and setting up chance after chance. It’s a shooting gallery. There’s not much traffic in front of the net, though, and not many sticks in the shooting lanes for deflections. James Reimer gets a look at every shot, and manages to come up with save after save, the rebounds trailing off to the side, or quickly cleared to the corner by Leafs defenders. Still, these are the Penguins, and Crosby, Malkin, Letang and Co. are racking up the scoring chances. The second goal hangs above the ice like a guillotine. At this point, a two-goal deficit would feel like four.
Then it happens. A whiffed shot from the left point drifts lazily onto the stick of Dion Phaneuf, who is busy keeping the slot in front of Reimer clear. Bang. The puck is off Phaneuf’s blade and up to a streaking Dave Bolland just across centre. Bang. Bolland turns on the jets and blows past a recovering Letang. Bang. He wires a slapshot past Marc-André Fleury.
It happens that quickly, and the game is tied 1–1. The execution is stayed. The Leafs will add three more unanswered goals and “thump” the Pens 4–1 despite being outshot 38–29, out Corsi-ed 58 to 42 percent and out Fenwick-ed 64 to 36 percent. Just another Saturday night at the ACC, where you can check your #fancystats at the door.
If you extrapolate hard enough, Toronto’s entire season—hell, the entire Randy Carlyle tenure—can be dragged out of that sequence: The Leafs barely touching the puck for long stretches, allowing the opponent to endlessly cycle in their zone, collapsing in a desperate shell around their overworked goalie and, of course, being badly outshot and out-chanced. And succeeding. The emotion most closely associated with a Leafs game these days is a sinking feeling that it’s all about to fall apart—a feeling that turns to utter joy when it does not.
Carlyle—who has been dismissive of analytics in the past—knows the Leafs’ style is not a recipe for long-term success. “We just don’t have the puck enough,” he lamented in mid-October, even as the Leafs sat alone atop their division with a 6-1 record. “We can’t just sit here and ignore it—we’re not going to.” But you can’t ignore results, either. Which leaves the hockey world trying to answer the most divisive question of the young 2013–14 season: Is this success just unsustainable luck, overdue to vanish and leave a mediocre-to-poor team scrambling to stay above .500? Or is there something else going on, a confluence of factors that nudges the Leafs’ defiance of analytic data away from the utterly inexplicable and toward the plausible? Well, you’ve got to be good to be lucky, and vice versa, and a deeper look at both sides of the debate offers a picture of a team at the crossroads of what this new breed of hockey statistics can quantify and where they still fall short of the reality on the ice. Or, as GM Dave Nonis succinctly put it during an appearance at the PrimeTime Sports Management Conference in Toronto in early November that seemed scripted specifically to irritate the analytics community, “Right now our Corsi stat sucks, but we’re winning hockey games.”
Numbers and ice hockey are strange bedfellows, especially in recent years—a hastily arranged marriage of order and chaos. Stats like Fenwick, Corsi and PDO can explain much of the big picture, without ever being tested by the madness of the moment. Those stats can be a balm that eases the pain of those 30-second snippets that make fans grip their remotes with white knuckles a few times per game, when bodies collide, pucks bounce over sticks and lead to odd-man rushes, and a goaltender’s glove darts out to snare a puck it probably had no business snaring. Those stats are a comfort to fans whose teams are outplaying the opponent but somehow, some way, losing. Crazy stuff happens, advanced statistics say, but don’t lose your head, because in the long run it evens out. It’ll be okay. And generally, they’re right.
There is a graphic that ran in this magazine—and was picked up in many places on the Internet—last spring (see p. 31). It illustrates the success of teams relative to their Fenwick close numbers—and it went viral precisely because of how closely it aligned with one’s own rough notions of which teams were the most successful over the past half-decade. A second graph charted the 2012–13 season with the playoffs approaching and the Leafs among the very worst in the league at 43.7, yet holding down the fifth seed in the East, the only team in six years to buck that Fenwick trend.
There is no denying that Toronto is a bad possession team—a club whose roster and style of play lends itself to being outshot on most nights and consistently puts it in a position where it must capitalize on a higher percentage of scoring chances than its opponent to come away with two points. All things being equal, these Maple Leafs will always be at a disadvantage there. At even strength with the score close after 15 games this season, the Leafs were 29th in hockey in both Corsi and Fenwick. When you remove the requirement that the score be close, they were dead last in both categories. They were also, one month in, tied for first place in the Atlantic division. Last season, they were last in Corsi and second-last in Fenwick. In PDO, the Leafs sat second in the NHL last year and fourth overall this season through October.
So yes, the case for the surprising success of the Leafs being little more than dumb luck isn’t hard to make—and you can find it stated angrily on the Internet anytime a Maple Leafs fan gets a little too proud of his team: Their Corsi sucks, their Fenwick is just as bad, their PDO is among the NHL’s highest—opposing shooters have scored 36 goals on 552 shots (6.5 percent) compared to the Leafs’ 47 goals on 392 shots (12 percent). Success with those numbers is impossible to sustain in the long run, statistics say. And, with the exception of the Leafs, recent results back those statistics up.
But those numbers are, well, numbers. They’re useful, but they’re not the final arbiter of success. The Leafs have been lucky, and some of that may well catch up with them, but their overall construction seems intended to defy the rules of Fenwick and Corsi. That’s not dismissing analytics in favour of trusting your sports gut or any other simple argument that ignores what those numbers offer. It’s an acknowledgement that, especially where the Leafs are concerned, there are a few other factors at play.
The relative quality of the shots heading toward Reimer and Jonathan Bernier does matter, even if we’re in the very early stages of being able to quantify just how much. Under Carlyle’s system, the Leafs keep opponents’ chances to the perimeter, making sure their goalies can see the puck. They are getting better-than-average goaltending from Reimer and Bernier—a hallmark of almost every team that punches above its analytic weight class and the downfall of many a Corsi- and Fenwick-approved darling in the playoffs. Ask any Red Wings fan how quickly running into a hot goalie can derail the seemingly inevitable coronation of a team that dominates in puck possession.
The Leafs also have a deadly transition game that leads to quality opportunities, and that in turn helps their elite players sport shooting percentages well above what would be considered sustainable. Their power play and penalty kill—which don’t factor in to Corsi and Fenwick five-on-five—are fifth and fourth in the NHL, respectively. They lead the league in short-handed goals (there’s that transition game again).
All these factors are capable of swinging a game in which a team is outshot and out-chanced. And it takes fewer of those swings than you might imagine to turn a mediocre team into a playoff club, and even fewer to swing a series once the post-season begins. Think of a stick exploding at the blueline, a defenceman tripping over his feet while backpedalling or the puck taking an awkward hop on a dump-in and ending up behind the goalie. If your above-average goaltender stops five of those chances at critical times and your league-high shooting percentage capitalizes on an extra five, and that swings five games, your 88-point team is a 98-point team, and the ninth seed is the fifth seed.
The Leafs are beating “better” teams, or, at least, winning a majority of games over teams that are outshooting them—and they’re being outshot so far this season by an obscene average of 10.7 shots per 60 minutes. They’ve been doing business this way, and succeeding, for more than a full-season stretch now under Carlyle (their first 82 games with him behind the bench resulted in 96 points, good for a playoff spot in every year in either conference since the lockout, except one—the West 2010–11). They can’t get much worse by the numbers and they’re better on the ice than they’ve been in a decade, and that somehow makes sense when you’re watching them play.
Carlyle, it’s worth noting, has a history of this sort of thing. Since winning the Stanley Cup with the Anaheim Ducks in 2007, he has coached five NHL seasons (four full and one shortened). In four of those five, his team’s Fenwick was below 50 percent, and in three of those seasons, his team made the playoffs. It’s debatable whether or not there is a formula here that is applicable to hockey teams at large, but it’s pretty clear that Carlyle’s teams aren’t judged fairly by Corsi and Fenwick. Even in 2007–08, with a fresh-off-a-Cup roster that included Scott Niedermayer, Chris Pronger, Teemu Selanne, Ryan Getzlaf and Corey Perry, the Ducks barely pushed their Fenwick above 50 percent, but their 102 points were the fourth-best in the NHL. When Carlyle took over the Maple Leafs, very little changed for the team as far as Corsi and Fenwick were concerned—the final iteration of Ron Wilson’s Leafs had a Fenwick of 46.39, while thus far under Carlyle they sit at about 44 percent—but the results are dramatically different.
Part of that is the addition of skilled players to the roster. Wilson was saddled with netminders like Vesa Toskala and J-S Giguère. Carlyle has the luxury of two talented goalies. Joffrey Lupul, James van Riemsdyk and Nazem Kadri are all offensively gifted players who thrive in the transition game and stand a better chance of capitalizing on limited opportunities than Clarke MacArthur, Nikolai Kulemin or Kris Versteeg. And elite scorers help you buck the shooting percentage trends. When van Riemsdyk was informed by a reporter that the average NHL shooting percentage was about eight percent at even strength (it’s around nine percent overall), his thoughts immediately turned to Phil Kessel—“We’ve got one of the guys with the best shots in the league, so he’s going to put it in more than eight percent of the time, I can guarantee you that.”
Now, van Riemsdyk will never be mistaken for someone who bothers with statistics—he referred to analytics in the same interview as “shooting percentage and all this crazy stuff”—but he’s right. Kessel has only one season—his rookie campaign—in eight years with a shooting percentage lower than 8.9 percent. He’s been at 12.4 percent or higher since the beginning of the 2011–12 campaign. Van Riemsdyk’s has also been above 12 percent in three of his past four seasons, with a career average of 11.1. Lupul has been under 10 percent just once in his past eight seasons and has a career shooting percentage of 11.7 percent. Kadri is shooting 14.8 percent over his first 114 NHL games. Off-season acquisition Bolland owns a career percentage of 14.8, and even the much-maligned Tyler Bozak carries a career 15.9 mark over 249 games. Alexander Ovechkin, for reference, is shooting 12.2 percent for his career.
On many other teams, the high percentages of those snipers would be dragged down by lesser players on the back end of the forward lines—but those six players account for more than 51 percent of the 392 shots the Leafs have taken this season. Kessel alone, shooting a cool 16.4 percent, accounts for 14 percent of the Leafs’ shots. And their career stats show that they’re unlikely to lose the ability to score at something approaching 12 percent or above—a number that allows for some regression from their hot start. Phaneuf, who leads the Leafs in shots by a defenceman, also owns a career percentage solidly above average for his position.
The Leafs currently sit second in the NHL in shooting percentage after leading the league last season and finishing sixth in 2011–12. The reason PDO uses shooting and save percentages to measure luck is because those numbers have a tendency to regress towards the mean over multiple seasons. But other numbers suggest that regression where shooting and save percentages are concerned—for the Maple Leafs, at least—may not be as drastic as analytics insist. Chris Boyle’s Shot Quality Project, presented by Sportsnet, which analyzes shot attempts and quantifies the value of them relative to the average save percentage on shots taken from that area, studied every shot taken by and against Toronto last season and determined that the Leafs’ unlikely percentages have some hard data behind them.
In the home-plate area in front of the net since the beginning of last season—where shooting percentage is highest—the Leafs have been outshot by their opponents 783–759, a seemingly small difference until you remember that they were outshot 2101–1656 over that period, or by an average of seven full shots per game. That means only about five percent of extra shots taken by the opposition were those high-percentage opportunities. The other shots came from the boards, or the blueline, and were more likely to be stopped—especially when your goalies are among the elite. Reimer finished eighth in the league in save percentage last season, and through 15 games both he (third) and Bernier (seventh) were among the best in the NHL this year.
Every one of these points throws a little more weight behind the idea that the Leafs can buck the analytics trend. It doesn’t mean that their success is guaranteed, because when you attempt to consistently win hockey games in which you lose the possession battle, your season will always balance on the edge of a knife.
For the Leafs to be successful, everything needs to go according to their uniquely precarious formula. Bernier and Reimer must continue to play at an above-average level. The snipers among their top-six forwards cannot afford to go cold. Their special-teams play must remain excellent, and any injury—like the severed tendon suffered by Bolland—has the potential to throw the whole thing into chaos. To beat the better teams in the league, the Leafs must play their game perfectly. When they do, the Penguins can dominate, yet slink off shaking their heads. When the Leafs fail to cash in on their chances, however, or collapse too deeply around their goaltender and surrender high-percentage shots, teams like the Blackhawks, Bruins and Canucks can make them look every bit the punchline they’ve mostly been in recent seasons.
It’s a delicate high-wire act, one that Carlyle has said the Leafs are sick of walking. But until they can fix their possession game, they’ll have to rely on the tools they do have. They’ll have to make their own luck. At least they know how to do it. —WITH FILES FROM CHRIS BOYLE