A few weeks before the season started, I wrote a piece about the 2013-14 Toronto Maple Leafs, because frankly, it’s hard to write about hockey statistics and avoid the subject for long. As outlined in this excellent piece at Grantland, the 2013 Leafs confounded the conventional wisdom that puck possession drives success, finishing fifth in the Eastern Conference last season despite wretched underlying numbers. As such, the optimism surrounding this year’s campaign in Toronto has fed into what many are calling a high-profile test case for hockey analytics. Will this season’s Leafs crash and burn like the 2011-12 Minnesota Wild did, or does Toronto know something that the numbers don’t show? If the Maple Leafs can win this season despite being consistently outshot, would the old-school defenders of grit and intangibles be vindicated? In some ways, of course, this conflict exists more in narrative than in fact: after a 4-1 win this week in which the team only managed 14 shots on goal, notable old-schoolers Don Cherry and Randy Carlyle both seemed to understand that that isn’t something they want to repeat. And Toronto’s team shooting these days is hardly as scorching-hot as it was last season. Still, with the Leafs 6-2-0 thus far, many are wondering openly if statistics are missing something when it comes to this team.
Earlier this week, sabermetrician Phil Birnbaum created a stir with this analysis, in which he suggested a method to Toronto’s madness. Noting that teams playing with the lead tend to both shoot less often and have a higher shooting percentage, Birnbaum posited that the Leafs are playing all the time as though they’re leading; that is, conceding a majority of shots while waiting for opponent mistakes that lead to high-percentage shot attempts. The implication, of course, is that Toronto’s sky-high 10.56% shooting in 2013 resulted (to at least some extent) from the team’s system of play, meaning that it isn’t necessarily going to regress significantly this season. A couple of analysts have already tackled this argument (see here and here), and others would agree that at least some of the variation in team shooting is non-random. For what it’s worth, I’m not sure Birnbaum’s argument squares with the data. As you can see in the table below (all data from stats.hockeyanalysis.com), Toronto’s 2013 shooting percentages don’t correlate in the expected way with their Corsi rates by game state. When trailing by one, the Leafs had both a sky-high shooting percentage and an above-average rate of shot attempts, which goes against the “selective shooting” argument. What’s more, they were one of the worst teams in the NHL at putting shots on goal when trailing, and were actually outshot when down by one; both of these suggest a team that struggled to generate consistent offense when it was needed, rather than a strong team playing opportunistic hockey. More generally, I don’t buy the notion of Toronto as a defense-first club playing for chances on the counterattack, because I don’t think you can separate good defending from effective play in the neutral zone. If you add up the Leafs’ Corsi For and Against rates per 60 as a measure of the pace of their games, only Carolina played hockey at a higher event rate than Toronto last season. Is there such a thing as a defense-first firewagon team?
In his defense, Birnbaum wisely avoids the conclusion that Toronto’s style is one that can win. Even if the Leafs somehow keep shooting at a high rate, the proportion of high-percentage shots matters less than the number of such shots per game relative to your opponent. Let’s say the Leafs play a team like San Jose, and get outshot 40-20. Of Toronto’s shots, two (10%) are high-quality (i.e., they have, say, a 50% chance of going in), eight (40%) are medium-quality (they have a 10% chance of going in), and the remaining 10 are low-quality (1% chance of going in). Now let’s say San Jose also gets two (5%) high-quality shots and eight (20%) medium-quality shots, with the remaining 30 being low-quality. The volume of low-quality shots ultimately gives the Sharks more expected goals, even though their shooting percentage on the night is much lower. Put another way, having a greater proportion of high-percentage shots doesn’t mean you’re taking more of such shots per game than your opponent.
Ultimately, though, if we want to understand how Toronto accomplished what they did in 2013, with an eye towards the current campaign, we need to look back at last season’s games in greater detail. Using the Extra Skater website, I compiled game-level information on even-strength shooting and goaltending, event rate, and Fenwick Close % for the Leafs. Much like my look back at the New Jersey Devils, a closer look at individual games gave me a better sense of exactly where the Leafs got somewhat lucky last season. Specifically:
- In three games against New Jersey, the Leafs were outshot 75-43 at even strength, and were out-Fenwick-Close’d 92-41. In one of the games, they failed to score an even-strength goal. They took six points from these games.
- In a March game versus Boston, they scored 3 goals on 13 even-strength shots, while giving up 33 shots and recording a 34.8% Fenwick Close. They won 3-2. Thanks, James Reimer.
- In a March game against Pittsburgh, Toronto was outshot 33-17 at even strength and crushed in Fenwick Close (27-8), and the otherwise solid Reimer recorded an 0.879 save percentage. Nonetheless, the Leafs got an overtime point.
- Reimer redeemed himself immensely in a mid-April game against Ottawa, which the Leafs won 4-1 despite being outshot 50-22 and recording a dismal 29.3% Fenwick Close. In another game against the Senators, Toronto won 5-4 while being outshot 43-28. In this game, the Leafs recorded only 7 Fenwick Close events, which is something considering that the two teams played at a pace of 143 Corsi events per 60 minutes.
- In late March, the Leafs gave up 42 shots and had an awful 35.9% Fenwick Close against the Florida Panthers, and shot an unremarkable 8.3% at even strength. Thanks to Ben Scrivens, they won 3-2.
- In February, the Flyers outshot Toronto 46-24 overall, and 36-18 at even strength. Thanks to 28% shooting and (again) Reimer, the Leafs won 5-2.
- In two games against Winnipeg, Toronto had a total Fenwick Close of just 35%, and were outshot 48-35 at evens. They took 3 points from these games.
So, for those of you keeping score at home, this was 20 points that the Leafs were at least somewhat fortunate to get. Had they gotten none of them, they would’ve finished 14th in the Eastern Conference. Had they only gotten half, they would’ve finished 13th. Had the Leafs missed just seven of these points, they would’ve missed the playoffs. Of course, one can surely go back to game data for any successful team and find instances of similar good fortune. The point is rather that luck did appear to make a big difference in getting Toronto into last season’s playoffs. Whether that happens again this season will have to be seen.