I have a reminder that keeps popping up to write another Corsi Report. I found the first one very interesting to do, and you all liked it, but then, that was back a few weeks when the season was young, and so were we. It was necessary to drop the Expected Goals models entirely, since the data was bad, and just look at Corsi and what what going on at that time.
The second Corsi Report hid under the name Occam’s Razor, and it’s time, I think, to take the temperature of the team again because they are nothing if not volatile this season.
We’ve talked about this a lot around here, in comments and in posts, and those conversations have been good ones. But at the end of it all, I’m not convinced by simple explanations. The Leafs have sacrificed offensive quality for shot share, is the correct phrasing for what is expressed often as: The coach has stifled the offence by making them play defence too much. Both the first phrase, which fits the actual results and the second, which doesn’t, are built on assumptions of cause and effect, and Occam is all about limiting the assumptions. Another assumption buried in there is that there is one simple cause. I resist that, and in fact, that’s how I feel in general: in a state of resistance to all the narratives so far. The assumptions they’re built on are fragile and will crumble when poked.
Okay, well, fine than, what the hell do I think? What’s my narrative of this season?
Facts before conjecture: We know the Leafs have travelled to a similar result in terms of shots and Expected Goals for years. What then changed this year to throw them off course? Before we look at the team and their play to answer that question, we need to understand the nature of reality itself. If we assume, as we often do when comparing one team to another, that all else is equal, we often get very wrong ideas about things.
I started questioning the nature of the hockey reality a few weeks ago, and I went down some really blind alleys. But I learned a few things along the way. One is that over the last five years, there is usually a moderate positive correlation between each team’s shot share and their results the previous year. The same is true of their Expected Goals %.
Note: In the post, the word shot never refers to Shots on Goal, but is always either all shots (Corsi), or usually Fenwick, which is unblocked shots. All stats are from time of writing and most include 20 games played. All data is from Evolving Hockey.
This year is not like other years. The relationship between each team’s Fenwick or Expected Goals % and their results last year is the lowest it’s been in this timeframe with 0.13 and 0.12 as the correlation coefficient, as opposed to numbers that ranged from 0.29 to 0.74 other years and were most commonly around 0.5. I ran these numbers on full seasons as well as the first 17-18 games for each season. The correlations were similar for the part seasons and the full seasons so this is not explained by there only being a few games in the books so far this year. Many teams are experiencing big changes from last year.
There’s another thing that’s different about this season. The differences between teams’ Fenwick %, or their ratio of unblocked shots for and against, and their Expected Goals % (The Fenwick number weighted for shot quality) is usually fairly small. That’s to be expected because the biggest part of Expected Goals is shot volume for and against. Most of the dramatic differences, the biggest swings in either direction over the five-year set of results, occur in this year. Many teams have a big gap in quality weighted shooting vs just quantity.
What is going on?
Teams change. They don’t tank, they aren’t bad, or they can’t afford more free agents or they change a coach, and from a confluence of reasons, this season is very different from last. We don’t remember things well, and our biggest impression of the past is last season, and if you feel like this year is very different, that something is weird, you’re not wrong. But it’s not just the Leafs. It’s most teams.
Some of this affects our perception. Just wait until the Leafs play the Canucks next month, boy will you be surprised. If you’re concerned because the Leafs have lost to a team you were sure was bad, maybe they aren’t anymore. But there’s more to it.
The first thing to understand is that while everyone is telling you the Leafs quality of offence is bad — they are one of those teams whose Expected Goals % at 47 is much lower than their Fenwick % at 51 — they are right, but the Leafs are not alone in this. And if you’ve been consoling yourself with the Leafs modest improvement in defence, maybe you shouldn’t, because almost everyone can claim that.
The Defensive Gravity Shot Up: We’re on Mercury Now
The boxplot shows 50% of the teams’ results in the box, with 25% above and the other 25% below. R takes out the outliers and makes them into circles. The thick line is the median each year.
Since this is Expected Goals Against, fewer is better. So as you can see, more than 75% of the teams this year are below last year’s median. And not only that, the middle 50% is a very tightly packed group of teams all allowing very similar xGA. The Leafs are in that box, on the high side of the median (the wrong side).
This isn’t unprecedented. The 2015-2016 season, when we were busy being bad, was similar. But the change from last year is large. The concentration in the middle is parity unlike anything we’ve seen in recent memory. If it feels some nights like every single team is about as good as the Leafs and can (and too often does) beat them, it’s because that’s true.
Not every team has declined in Expected Goals Against. Not quite (the bars on the left show improvement since fewer is better).
The Rangers got a lot worse. The Habs and oddly, the Bruins and Tampa did as well. And the Leafs are a bit below average for the decline, nothing like the Ducks. Gee... what could be going on there that they got better defensively?
If the shots against are down all over, then offence must be too.
It’s tough to get a goal out there these days
This is less compressed than the Expected Goals Against, but it’s still clear that there are fewer teams riding high and getting super-charged offence. This is broadly similar to 2017-2018, although the median is a lot lower. The Leafs are below the median, but still in the box that holds the middle 50%.
At the team level, we can see a few more teams bucking the trend:
Again the Ducks have this huge improvement... hmmmm. San Jose fans must be way more upset than us, but it’s clear that the Leafs’ decline in offence by this measure is not just the global shortage of goals. It very definitely is a change in the way the team plays, but the true nature of the change is magnified if you just look at the Leafs in isolation year-over-year and are aghast.
For most teams that are seeing either a rise in xGF or a bigger drop than average, there’s usually a reason. Either a new coach or a changed roster (poor Columbus). But if even the powerhouse Carolina Hurricanes are doing worse, as are the Pittsburgh Penguins, making that two of the league leaders in Expected Goals % showing a drop, it should not surprise us that the Leafs, with their roster changes, are at least suffering the average amount of drop.
The question then becomes, why the above average drop?
Can anyone get a decent shot off here?
Let’s have a look at the scale of the Leafs players drop in On-Ice Expected Goals this year over last. This is not individual offence, but what is produced by everyone when that player is on the ice.
Rome was not built in a day, nor was this article, so this segment includes 21 games of data. I’m using the relative to teammate measure here (a weighted relative measure) to get some idea of who is genuinely improving or declining in effectiveness.
Offensively, where almost everyone is worse, as we should expect:
And defensively, where better is the bar on the left now, but the results are all over the place.
Note: Petan is being compared just to his larger sample of Jets games, Muzzin is just set against his Toronto play. This is ordered by who was best at Expected Goals For last year. And in the xGF chart more is improvement, in the xGA chart, less is better.
The Matthews line has improved against the tide of the league’s changes, and perhaps their own weaknesses. The Tavares line is the problem. (Zach Hyman has one game, and some of the other depth players have very low minutes, so don’t concern yourself with them as much.)
The fourth line is getting a much, much harder job defensively, and their results are what they are. But the overall quality of the depth in terms of their ability to generate offence in the NHL as it is this season has declined. Many of the Leafs’ best players are worse defensively.
More detail on who is individually getting quality shots is a project for another day. But it’s widely understood that the Leafs are seeing defencemen shoot more than they were and more shots are coming from the points, while fewer are coming from the net-front.
Putting this into a story
Here’s what I think happened. I think the Leafs, and I mean Kyle Dubas and Mike Babcock and everyone else together, set out to deal with a very real problem last summer. They didn’t have Nazem Kadri and Jake Gardiner anymore. Both of those players got good shot share, contributed to the high pace of Expected Goals For and were good defensively. Tyson Barrie is not. Cody Ceci is not. That is a fact that the Leafs had to be aware of.
Kyle Dubas decided to lean in on offensive prowess over defensive. Partly this was circumstances, but partly choice. Getting Tyson Barrie when he already had Morgan Rielly was like going to a bar and ordering a beer with a beer chaser. To deal with this reality, they chose to make some system changes that not only suit their team, but would hopefully result in a better shot share. Because less time spent in the defensive zone means less time with the players struggling with their weaknesses.
And this worked! It looks like the choice was to carry the puck through the neutral zone more and maybe to shoot in different ways. It’s impossible to tell if the change in shooting was intentional or just a side effect of personnel and the speed with which players are now entering the zone (slower). But there is some evidence that rush chances are down, which impacts the overall Expected Goals For.
But this scheme to maximize the overall skillset of the team slammed head on into the new gravity in the NHL where it’s harder to get into scoring position, and harder to get shots. The competition is better. And we like to say that the effects of competition washes out over time, but the wash water has changed. It’s made of acid now, and it eats away at the offence of every team.
For the Leafs to be maintaining their Expected Goals For pace of last season, they would have to actually play better offensively by a significant amount. Can we call that a reasonable ask given the roster? Add in some goalie trouble, a power play that didn’t work, and once you start playing from behind all the time, your shot quality gets worse, not better. This became a spiral of losing very quickly.
Now what, though? If the Leafs aren’t simply being stifled by a coach who hates fun (if Mike Babcock is stifling the offence, he’s coaching a hell of a lot of teams) then what’s the response to get the team on a path that hits a balance between offence and defence that works out in the Leafs’ favour?
It looks like the easiest thing to do in the NHL right now is to get better defensively, to tighten up the execution at the very least, but for the Leafs, surely a return to a run-and-gun offence plays more to their skills. Which is the right way to turn the dial? Do you go with Dubas and get fries with a side of fries, or do you go with Babcock and point out that a little grind of pepper on those fries might make them taste better?
That’s a very good question, as Mike Babcock likes to say when he’s not going to answer.
I think this problem will fix itself a little, but only a little. Let’s look at the Leafs as part of the league:
This is the Threat model, a simpler xG model to what is used above, and it’s less harsh on the Leafs in terms of shot quality effects. It gives us a picture of the path the Leafs are on. They are sitting around the 50-50 line, and to get better, they either have to move to the right or up. Or both.
If the Tavares line starts playing with the sorts of defensive improvements they should have just by virtue of playing in the NHL in this season, and if they can sort out their offensive issues (which revolve very heavily around Tyson Barrie and Jake Muzzin, who are both shooting the puck a lot) then the team might do both.
The shots are there! The zone time is there! The Leafs are way, way to the right of almost the entire NHL by just shots for. And the single, most overwhelming component of Expected Goals of any kind is volume of shots. They very obviously need to improve defensively somehow because they allow too much by any measure. It’s trite and just false to suggest that reduced offensive quality is the only problem.
I see the system this team began this season with as an attempt to tune the Leafs just a little that unfortunately met a situation that distorted the results out of the acceptable range. A re-tune is very much needed because the plan didn’t work. There is no question about that. But it wasn’t unreasonable in theory, and the results look worse than the system changes can account for.
It’s really difficult to see game by game how any re-tune is going. We had a couple of wide-open goalfests that the team lost, and we had a very good, balanced outing against the Bruins that they also lost. And then we had total chaos in Pittsburgh.
Judging the process while you’re hip deep in changing it is tough, but that’s why Dubas and Babcock get to sort this out while drawing their nice, large pay cheques, and I just get to watch. But this new high gravity NHL is not going away, so they need to deal with reality as it is, and not how they wanted it to be when they drew up their plans.
Comment Markdown
Inline Styles
Bold: **Text**
Italics: *Text*
Both: ***Text***
Strikethrough: ~~Text~~
Code: `Text` used as sarcasm font at PPP
Spoiler: !!Text!!