UPDATE, October 15: We now have reason to believe that the Expected Goals and shot plots used in this article are flawed. It’s not quite true to say that everything is now fine, but the conclusions here are not all supported by clearer understanding of the data and how it’s recorded.
The Toronto Maple Leafs played their worst game on Thursday night since they got rolled by the Tampa Bay Lightning in October three years ago. I vividly remember that game, because I re-watched it in slo-mo with the isolation cam on Frederik Andersen. It was actually a festival of such poor play, it’s hard to imagine last night was as bad — I vividly remember that 2016 Leafs team consistently left Victor Hedman open to float around with a clear shooting lane and the opportunity to dish it to the also-open Steven Stamkos. Last night wasn’t as bad. Or at least it was bad in different ways. And Andersen likely was more culpable last night than three years ago during his famous “October Freddie” period.
What obviously went wrong last night was Kasperi Kapanen. He’s been crappy Kappy all season, and that stick throwing incident was just the thing that opened the floodgates and gave everyone permission to finally say he’s been bad. He’s been terrible, and leaving him on John Tavares’s line is stupid and anyone who does it should be fir— wait. Are we sure about that? That all sounds a bit emotional to me. Let’s take a look at the facts on the case.
How is everyone doing?
First, a 10,000 foot overview of the team’s play so far at five-on-five:
On-Ice Threat: Micah Blake McCurdy has described his Threat model as a simple Expected Goals model. It is simply all unblocked shots weighted for the quality of the shot location. It’s just sophisticated enough to stop someone who is on the ice for a lot of shots from poor locations from showing up better (or worse) than they are. This is a really good view of overall pace of play of each player and how far they are from the team average (the TOR at the intersection of the blue lines, covered over by several players).
Ignoring the low-minute depth players, we can see that Auston Matthews is in a class of his own so far this year. This means that both William Nylander and Andreas Johnsson have had some shifts away from him that were worse or he’s had a great shift with some other players.
The top four defenders and most of the rest of the top nine forwards are clustered around the average, and no one stands out dramatically. John Tavares seems to be getting bad defensive results, and Mitch Marner seems to be getting bad offensive results and Kapanen is strung on a string between them.
Before we go any further, we should look again at all these players up against our expectations for them. I expect (hope) Cody Ceci just to not be a net negative impact player. I expect Morgan Rielly and Tyson Barrie to drive offence, and for Jake Muzzin to be good defensively. I expect the Matthews line, now with Nylander where he belongs, to be elite all year while they play against top competition a bet less than Tavares does. I expect the Kerfoot line to be Leafs average and for the Tavares line to be the backbone of the team, good at everything, playing the top lines, succeeding offensively, and good (graded on a Leafs scale) defensively.
- Ceci: so far, so meh
- Rielly: what the hell is going on?
- Barrie: before the Tampa game, he was lights out offensively, but that game killed his stats
- Muzzin: okay but not great
- Matthews line: with the caveat that Andreas Johnsson will always be the worst player on that line, they look great.
- Kerfoot line: Kerfoot himself is better than his wingers, but they’ve been fine
- Tavares line: uh.... you guys were supposed to regress in shooting luck, not playing ability!
Is the Tavares line bad?
It’s time to look at a less simple model of Expected Goals and see what’s to be seen. I’m utterly ignoring the power play for this article as well.
Moving to Natural Stat Trick, and looking at Score and Venue Adjusted five-on-five Corsi For %, the Leafs are ninth in the NHL with 53%. That’s excellent. By Expected Goals %, they are 16th at 49%. That’s average, and that’s a problem. The Leafs should be (small E) expected to be running on offensive skill and to be better than their shot share when weighted for quality. No team plays their most skilled players all the time, so the depth will eat into that, but the Leafs have a lot of skill.
As of October 11, 2019
|Line or Pairing||TOI||CF%||xGF%|
|Line or Pairing||TOI||CF%||xGF%|
The problem seems obvious. It’s the Tavares line. No wait, it’s Rielly-Ceci. What if it’s both? When the Tavares line is on the ice with Rielly-Ceci (24:43 minutes) they are so terrible, it’s hard to imagine. Their xGF% is 36%. But before you go ah ha, and turn on Ceci, because, of course, it can’t be Rielly, when the Tavares line is with any other defensive pair, they are better, but still show this terrible drop from simple Corsi to Expect Goals.
They’re bad defensively, aren’t they?
Now we need to know which side of the ice this is coming from. As far as defenders go, the Rielly-Ceci pair has the best Expected Goals Against per 60 minutes at 2.2. Muzzin-Barrie is at 2.49. These guys are still Leafs, and the Leafs play high-risk hockey, so that grades out as 57th and 77th out of the 97 pairs that have played at least 20 minutes.
The reward of high-risk hockey is supposed to come from heavy offensive pace, and Muzzin-Barrie are 20th in the NHL at Expected Goals For per 60 minutes. (I shouldn’t mention this, but it’s sitting right there... Jake Gardiner and Haydn Fleury are first.) Rielly-Ceci are 68th, and that’s a problem.
Expected Goals For pace comes from two unequal sources. Shots is the biggest part of it. Quality of shots is the smallest. The Leafs should be ruling the NHL in Corsi For per 60 minutes, and they almost are. They just need to dethrone the Flyers (Alain Vigneault likes high-risk too), and they’ll be first at 65 Corsi For per 60 minutes. But it’s Matthews who is driving the Leafs offence right now. Tavares is not really bad, but he’s not where he should be.
We’ve come full circle. Look at the first chart again, just at the horizontal axis, and It’s Marner and Kapanen who are lagging, but also, particularly, Rielly.
Something is blocking the Leafs offence
Something funny is happening on the way to the opposing net. Expected Goals are based on Fenwick (unblocked shots) because they have to be. But Corsi gives a good overview of who has the puck the most. But on the way from Corsi to Expected Goals, the Leafs have some odd Fenwick results. The Tavares line is showing a very large drop in offensive pace (CF/60 to FF/60) compared to the other lines on the ice. A lot of their shots are blocked. Morgan Rielly’s FF/60 is so low, he’s almost in Frederik Gauthier territory.
The other side of the Rielly-Ceci coin is that their Fenwick Against is very good, and it’s their reduction in their very bad rate of shots against to fewer getting through to the net that’s helping their Expected Goals Against.
“The problem” here is on the offensive side of the ice, and that problem of high blocking rates and low unblocked shots rate is made worse by some of the best offensive players on the team not spending enough time in the offensive zone because they’re not exiting the defensive zone very well. Just because you defended well while you were there, doesn’t mean the game is now 10 minutes longer, and you get more chances to get in the offensive zone.
This is hazy territory to be in. Shot blocking rates are heavily situational and dependant on opponents, and the Leafs did play the St. Louis Blues, so the seeming lack of offence from Tavares and Rielly might just wash away in a larger sample of games, but it can also mean there’s a problem with shot location. And there is:
This is the Leafs last year: (Red is good, close to the net is best.)
Leafs this year so far:
Now, note the threat. The Leafs are grading out, as a team, at below league average offensively. And yet, if you shoved that red blob up towards the net ten feet, you’re pretty much good. Okay, maybe 20 feet.
Defensively, as a team, the Leafs are league average according to Threat, which, hell, we should have a damn parade for that. Mike Babcock said after Friday’s practice that he likes their defensive numbers, although he never elaborates about what they measure. He’s not blowing smoke, though.
I don’t buy into the story that the coaches stifle offence by making the poor little simple-minded players worried too much about how bad they play in their own end, however. The key to offence begins in getting out of that defensive zone. And lest you think the Leafs are suddenly a brand new style of team, they have allowed the second most Corsi Against in the NHL so far this year. They’re still a super high event team, don’t worry.
Is anyone shooting well?
I’m switching to Evolving Hockey for this next bit, because I like their Expected Goals model a little bit more. But I want to touch on individual shooting.
Tavares and Marner are shooting exactly at the pace they did last year, except their Individual Expected Goals is very low. Kapanen, in a different role, is shooting less, but Nylander and Johnsson are as well, deferring to the wildly successful Mr. October.
Rates of shooting aren’t the problem for the Tavares line. John Tavares’s personal shot location is much more worrying than anything anyone else is doing. His previous season is a masterclass in quality shots.
So, back to what it seemed from the very first chart, which I keep swinging back to. Morgan Rielly and John Tavares are not the offensively dominating players they should be so far. Mitch Marner just seems a little dull, and Kasperi Kapanen is not the straw stirring the drink or the one breaking the camel’s back. He’s doing okay in a role he’s not really suited for.
I don’t think either Rielly or Tavares are suddenly bad. I think they’re all getting used to some new teammates, and they’ll settle in. But Zach Hyman isn’t going to make this situation worse, that’s for sure.
Nothing anyone has done so far indicates that there’s a really good reason to mix up the lines. Ilya Mikheyev is not performing better than Kapanen, Trevor Moore has been the worst player on the third line, and the only other option is to swap Kapanen for Nylander and see if he still remembers his two week trial run as a left wing from the World Championships. But no one is breaking up that Matthews line. Not when it works so well.
Zach Hyman practiced in a regular jersey for the first time on Friday, and he’s eligible to come off LTIR in late October. Until then, it really up to Rielly and Tavares to push themselves to their usual standard of play. They’ll get there.
So, no. Kappy isn’t crappy, he just needs to buckle his helmet on tighter and keep doing his best.