The Toronto Maple Leafs have played seven games, have four wins, two losses and one loss of a coinflip in the shootout for nine points and are, well... what are they? With the unreliability of shot location data at the moment, I want to take it as an opportunity to simplify things, and go back to the old fashioned methods of constructing narratives by hand. I’m not going too far, here. I haven’t ripped out my oven and put in a wood stove, and I’m not going to start using plus/minus and boxcars to measure the team either, but there’s nothing at all wrong with measuring the team by shots.
Not that kind of shots.
Corsi! We’re bringing Corsi back.
For the youngsters who don’t remember back to 2012 or so, Corsi is the simplest measure of shots you can have in hockey. It’s all of them. All shots that become goals, saves, misses or are blocked on the way to the net are Corsi For or CF when they’re yours and CA when they’re the other team’s.
Why use shots? (Skip to the good stuff if you know all this.) Hockey has less than three goals per team per game on average. Last year, teams produced 2.81 goals each per game, which was the highest that’s been since 2005-2006, which you might recall was the first year that some attempt was made to call obstruction penalties. Because there’s so few goals in a game, a weird bounce off a stanchion or one moment’s inattention by someone on the ice, and a goal can go in that decides the game.
That kind of fluke doesn’t happen all the time, but it’s often enough that you need some event that occurs in the game more often than goals to measure how well a team plays. Shots are that event. That’s all there is to it. There’s charts and graphs and analysis about how repeatable and predictable using Corsi is, and there’s ways to make the raw shot figures more valuable, but if all you ever look at is unadjusted Corsi, you’ve used a good ruler to measure the team’s play. If all you look at is goals, you haven’t.
Corsi is nearly always used with the five-on-five play on the ice, or even-strength. I stopped looking at even-strength numbers when three-on-three overtime was implemented because I think it’s fundamentally a different thing from the actual game, and can be ignored as anything but entertainment.
Why five-on-five and not the whole game? In a recent Back to Excited, Fulemin did this bit where he mockingly described how a classic hockey nerd feels about the power play. It was something like how five-on-five is the real game, and the power play is just some other thing entirely that has nothing to do with actual hockey, and I was nodding along and saying, “Yeah, it isn’t hockey. Plus, I feel attacked.” But leaving aside the extreme nerds who forget it exists, the power play shouldn’t be considered in the same way as the rest of the game.
How important power play skill and results are is something that likely varies from team to team. If you don’t have the players to execute it well, you have to win the game at five-on-five. But the reality is, whether people like to hear it or not, every team that wins is doing it at five-on-five. And you can best measure that five-on-five ability by Corsi. The most recent not-terribly-gifted five-on-five team to win it all was the Washington Capitals, and since no one can set out to have Ovechkin on the power play, the rest of the NHL has to play the rest of the game, 50 minutes or so of every night better than most of their opponents most of the time in order to win.
Varieties of Corsi
I like two types of Corsi, and I’m done prefacing everything with five-on-five, you can safely assume that’s what I mean. I also never use the word shot to mean shot on goal. Looking at that subset of shots, as television broadcasts still mostly do, is a waste of time when there is Corsi data.
Score and Venue Adjusted Corsi: is the adjusted shot numbers produced by Natural Stat Trick. The underlying concept is simple: When a team is trailing they shoot more. Home or away status has a small effect on this, enough to include in the adjustment. Adjusted Corsi simply says that those shots you get when trailing are “worth less” since they’re easier to come by. The adjustment makes Corsi more predictive and makes it a more realistic description of a game.
Rel TM Corsi: Before there were expected goals models, there was this improved form of relative Corsi. Relative simply means each player’s Corsi is considered in the context of his team’s average. A player might be 45% on his season, but if his team average is 50% or 55% or 45%, that player result means different things. Realative Corsi is expressed as the distance from the average. positive CF and negative CA numbers are what you want to see.
The Rel TM is a calculated amount produced by Evolving Hockey using a simple formula that gives a broad representation of how well a player does relative to how his teammates perform without him. It’s a more complex WOWY calculation, and really, don’t ever use a WOWY when this is available instead.
The Good Stuff
Enough of that, let’s talk Leafs.
The Leafs kicked the Flyers off the top of the Corsi list and now reign the undisputed number one team with 56% (I have little patience for fine grained distinctions to multiple decimal places and I tend to round things off).
Last year, the top team, the Sharks, finished at 55%. The year before, the Hurricanes finished at 54%, so chances are the Leafs will get worse not better. But so far, they are the tops.
The got to the top by having the highest pace offence in the NHL (where they also just dethroned the Flyers, who must have had a bad game). The Leafs are producing 67 CF per 60 minutes. Over the last two years combined, the top team, the Hurricanes, produced 63.
Pace of shots is made up of two things. How much you actually shoot, and how much time you have in the offensive zone. This year so far, the Leafs have dramatically improved the second thing (as implied by their CF%).
The Leafs pace of CA per 60 minutes has been bad for years, sixth worst overall in the last two years combined at 58. This year, so far it has been excellent at 53, which is seventh best.
This is an astonishing change, and has been the driver of most of the success on the ice for the team so far this year. Will it last? Well, seven games can have a lot of random variance in it: strength of opponents, injuries on other teams, rest issues, etc. But it’s still a very good early sign of a successful system that’s being executed well.
Corsi tends to be a bit wild with big swings in results early on, and then it tightens up as teams play a more representative sample of games. Early games are often a bit wild in structure too, which has an effect. But 53 CA/60 was third best over the last two seasons (St. Louis Blues).
For the players this time, I set an arbitrary minimum of 40 minutes played. This keeps Justin Holl but cuts the fourth liners who have rotated in and out. Those small bits of games are just not worth the time spent.
Who is driving that offence? This is the RelTM CF/60 leader board:
- William Nylander: 8.88 (I am not making this number up)
- Auston Matthews: 7.02
- Cody Ceci: 4.5
- Andreas Johnsson: 2.53
- Tyson Barrie: 1.41
A theme has developed there. Matthews has spent 50 of his 99 minutes with Barrie and only 34 with Rielly, so that five-man unit is driving the offensive train for the Leafs so far. Tavares is also positive, just, and Alexander Kerfoot seems to be the team average Maple Leaf so far this year with -0.2.
The worst offensive forward is, of course, Frederik Gauthier, with Trevor Moore and Ilya Mikheyev not much better. They’ve both been fun to watch, but try to be realistic about how well they’ve played. Kerfoot is the real star of that line, even if he’s harder to notice.
The leaders in limiting shots, the RelTM CA/60 stars are:
- Auston Matthews: -7.57
- Mitch Marner: -6.16
- Tyson Barrie: -5.71
- Justin Holl: -4.64
- Rasmus Sandin: -2.68
The presence of Holl and Sandin, edging out Nylander, Kerfoot and Muzzin (who sits at about team average) is likely down to very careful usage so far. Holl has performed well in his role, however.
The worst player for shots against is Moore, with Ceci, Gauthier and Morgan Rielly, right behind him.
How the Leafs got their results
Gauthier gets a few extra shots against to his name from taking Faceoff-Getoff shifts, but there is still a very large gap in the pace of shots against between the fourth line, most of the third line as well as the Rielly-Ceci pair and the rest of the team including Kerfoot.
It’s normal for Morgan Rielly to be bad at shots against. The gap between him and Ceci in shots for could be down to the coaching staff using Ceci more when the team is leading by more than one goal. When a team is up by two goals or more, the offensive skills of the players are least important, and their defensive shortcomings are less likely to harm you, and that’s when the Leafs play Ceci in place of Barrie.
Matthews is having a glorious year so far, in part because of Nylander, and also because of the defence pair he sees the most time with. Barrie is nearly Nylander-class at moving the puck out of the defensive zone and into the offensive zone, and his usual defensive shortcomings are deep in hiding so far. If they can keep this up, they will continue to dominate the NHL.
Matthews is succeeding mostly because of who he plays with, but beyond that, the biggest factor is Alexander Kerfoot. Kerfoot is taking a big share of the minutes against top lines so far, much more than Nazem Kadri ever did. He’s holding his competition to even, and allowing Matthews a clear run to the opposition net by rarely facing a top line.
Kerfoot’s stats look very mediocre, but he is a big part of the Leafs hot start.
Who’s too hot not to cool down?
Matthews is getting some shooting luck, but there’s a host of players who have zero luck so far at five-on-five, so it’s going to be pretty painless when he cools off. The Kerfoot line likely has more points than they’ve earned, but they’ve been doing their jobs so well that also doesn’t really matter much.
But underlying all of this is the man so cold, you wonder if a new, heavy woollen suit isn’t in order to warm him up. Frederick Andersen has been terrible. To avoid the Expected Goals mess right now, I looked at Andersen by Goals Saved Above Average: GSAA from NST.
This is all shots on goal faced by the goalie with a league average save % applied and then you get the total goals that goalie has saved above that number.
Andersen is -2. He helped himself out last night some, but of the 34 goalies who have played at least 150 minutes (or three games), Andersen is ninth worst by GSAA/60.
He’s going to get a lot better than that, and when he does, the Leafs are poised to just keep rolling and maintaining a standings spot near the top of the league. They’ll be sharing it with other teams in the Atlantic, though, so that much hasn’t changed. Even if Tampa and Buffalo have traded places so far this season.
That’s the early Corsi Report. This was fun, so expect to see it again at some point around game 15.
Oh, wait... I forgot the power play. The Leafs have 38 minutes on the power play so far and 37 on the penalty kill. That’s too little to learn much from.