When you look at the standings over the last few years, the Leafs have progressed in a very linear way. They were bad three years ago, average two years ago, good last year, and are great this year. By the standings, they’ve progressed the way that we like to imagine all young teams do. Generally speaking, the media and fan coverage of the team has reflected this.
However, there’s been a notable contingent of Leafs fans who are not quite satisfied with the team, largely due to the team’s middling shot and expected goal share averages throughout this time. The goal results have gotten better, but over the last few seasons, the Leafs have been remarkably consistent in how they control play at 5v5.
Leafs 5v5 shot metrics (via Corsica)
|CF%||50.59% (15th)||50.88% (9th)||50.71% (14th)||51.00% (11th)|
|xGF%||49.26% (17th)||50.89% (11th)||51.07% (13th)||51.36% (10th)|
The ranks are unimportant here... in the middle of the league, the differences between teams are very marginal. The consistency of the Leafs is immediately obvious. An important point to acknowledge (though I won’t go any deeper into it) is that not all 50% xGF% teams are made equal. All things being equal, a team that produces higher quality chances to arrive at the same average xGF% as a team who produces more lower quality chances is more likely to win a game. In this sense, comparing teams by xGF% is a bit short-sighted... we are effectively comparing two probability distributions by looking at a function of their means. Regardless, it suffices to do so for the purposes of this article.
Back to the main point. The largest takeaway in hockey analytics from 2010 to 2014 or so was based around the fact that controlling shot share at 5v5 did a better job of explaining how teams would do in the future (in terms of goals) than looking at how teams did in the past. That’s why we care about Corsi. Since 2014, we’ve also been able to determine the existence of shot quality in an offensive and defensive sense at the team level and better quantify it, while doing the same for sustainable shooting talent. However, the core tenets of what we learned in that era still largely hold. It’s not a guarantee of success, but, given the choice, you’d rather get more and better chances to score than your opponent (setting aside score effects).
Teams that tend to go deep in the postseason, with a few notable exceptions, are teams that do this (though it’s worth noting that recent champions like the 2017/2018 Capitals and the 2016/2017 Penguins were not elite shot share or expected goal share teams). The Leafs have been average at this over the past few seasons. That’s where the concern arises from on the part of Leafs fans.
This has led to criticism of Mike Babcock’s ‘system’, the Leafs’ reliance on the stretch pass, and general confusion on how a team with as much talent as the Leafs can so regularly look like they met each other five minutes before puck drop. However, despite these criticisms and the Leafs’ middling shot metrics, their 5v5 goal differential has steadily improved over the past few years.
Leafs 5v5 Goal Difference
|5v5 Goal Difference||-40||8||29||69 (projected)|
At 5v5, the Leafs have gotten better results than their shot metrics would imply over the last couple years. This is the result of getting strong goaltending from Frederik Andersen and converting on their 5v5 shots at a higher than average rate. The former is about as reliable as any goaltender this side of John Gibson. The stability of the latter is the subject of much debate. Certainly, the Leafs have a couple players who tend to outshoot expectations (most notably Auston Matthews) and a couple players who may inflate teammates shooting percentages. However, it’s fair to say that relying on percentages going your way can feel a little tenuous in a way that outshooting and out-chancing teams doesn’t, especially come playoff time. In the playoffs, you’re generally also facing teams that are above average in terms of goaltending and shooting talent. Most of the bad goaltenders don’t make the playoffs! And even a shooter like Auston Matthews can go cold over the course of a series, as we saw last spring.
However, over the last nine games*, the Leafs’ ability to carry play at 5v5 has gone up a notch. From the start of the season to game 28, the Leafs had a score adjusted CF% of 50.36%, ranking 13th in the league, right in line with where they have been over the past few years. In games 29-37, the Leafs had a 54.57 CF%, ranking 4th in the league. A similar pattern persists when you look at expected goals.
* I’ll explain why I chose nine games shortly.
The obvious caveat is... it’s been nine games. Many teams have strong nine game stretches. Montreal is 2nd in CF% over the same time span. Philadelphia and St. Louis are 5th and 6th in xG%, and I’m pretty sure both those fanbases are closer to setting fires than planning parades. Hockey is a random sport, and the gaps between teams are relatively small. The difference between a 50% CF% team and a 55% CF% team is generally less than 10 shot attempts per game. It’s tempting to chalk this up to variance, and leave it at that.
Another thing to note is that this has been a soft stretch of the season in terms of quality of competition. Over the last nine games, the Leafs have faced Detroit twice, Florida twice, Carolina, New Jersey, and the New York Rangers. While tougher games against playoff teams like Tampa Bay and Boston have been mixed in, a lot of the teams mentioned above are fairly low in the standings. However, a few of these teams are actually decent at controlling 5v5 play, which is what we’re actually interested in measuring for the purposes of assessing whether the Leafs’ improved shot share is a mirage. Carolina is famously elite at driving play, as are Boston and Tampa Bay. The Panthers and Devils are mediocre at doing so, while the Rangers and Red Wings are bad. On the whole, it’s not a murderer’s row, but it’s also not entirely cupcakes.
Others have argued that the Leafs have trailed more in this time span, and that juices their shot share numbers, even once score-adjusted. As far as I can tell, there’s no hard evidence regarding whether some teams are more or less susceptible to score effects than others. By definition, score-adjustments are based on league average shot rates in each score state, and without more solid reasoning behind why a team like the Leafs would be disproportionately impacted, I don’t really consider this a valid argument. In any case, the Leafs are 6th in CF% when the score is tied over these nine games.
Nine games also seems like a very arbitrary endpoint. However, there is a hockey reason as to why the Leafs play may have changed nine games ago.
William Nylander has now been back for nine games, and while he hasn’t lit the world on fire, the games following his return have been among the best of the season for the Leafs.
His return has given the Leafs three strong centre—right wing pairings. Even when not at his best, Nylander puts pressure on the opposing defense with his reputation and skill set, and allows all of the Leafs’ top three centres a player who is capable of bending the defense and playing off them. Without Nylander, Kadri has puttered around with about the same success as the rest of the team in terms of shot share (around 50%). With him, they’re operating at a ridiculous 60% CF% (the same pattern holds if you look at unblocked shots, scoring chances, or expected goals). They’re too good to not beat up on middling players on other teams, even when Nylander has clearly been finding his legs.
Of course, Nylander is not the sole cause of this uptick, or even particularly responsible for it. The Tavares/Marner duo has continued their domination of top lines and pairings, and Matthews/Kapanen have done the same. In almost any configuration of those three centres and those three right wingers, the Leafs can expect to get elite results out of their top nine. There’s room for optimization beyond that... Andreas Johnsson looks like he’s found a home on the Tavares/Marner line. There might be a LW rental worth having to upgrade that side (Marcus Johansson or Gustav Nyquist are potentially interesting choices). Or maybe Trevor Moore can take that spot and make it his own after a solid debut. The injuries to Tyler Ennis and Zach Hyman hurt a little, but both are depth pieces who are fairly replaceable, all things considered.
Of course, the small sample size spectre hangs over everything we’ve seen so far. However, it can only be considered encouraging that with their full lineup (or close to it), the Leafs have performed well on the spreadsheet and the scoresheet.
As Ian Tulloch (among others) has pointed out, the Leafs can reasonably expect to sustain slightly above average shooting and goaltending. The Leafs have ridden those percentages to where they currently are in the standings. Improving their ability to drive play at 5v5 solidifies their position in the league table, increases their margin of error in winning the shooting percentage battle, and narrows the gap between them and Tampa. In short, it makes the Leafs more of a contender than they were. Long may it continue.