The Toronto Maple Leafs have played 21 games, so that means it’s time to look below the surface and deeper than the standings to see how they are doing so far.

Disclaimers

This was written before game 22, and includes data as of November 20.  All numbers are either the Score and Venue Adjusted five-on-five from Natural Stat Trick or, in the case of anything with expected in its name, the unadjusted five-on-five numbers from Corsica Hockey.  Twenty-one games of Corsi Data has been shown to be close to the peak of predictive value for future results both in Corsi itself and goals and wins. Expected Goals data is used for descriptive purposes only.

Definitions

Corsi: All shots, whether they hit the net, miss, or are blocked. sometimes called shot attempts. I will use the word shot or shot differential to mean Corsi. I am never talking about Shots on Goal.

Fenwick: All shots that aren’t blocked.

Expected Goals: A shot quality measure calculated by weighting Fenwick by shot location and type, as well as other factors, depending on the model used. I am using Corsica’s.  xGF/60, for example is the number of goals per 60 minutes you should expect on average from shots of the type taken.

Fenwick Save and Shooting Percentage: Just like the Sv% and Sh% you’re used to, only taken as a percentage of all unblocked shots instead of Shots on Goal.

Corsi Then and Now

I’ll begin with the basics and see what we can get from it. This year I am comparing the first quarter of this season to the tank year and the barely made the playoffs year that followed.  The old days before that are gone for good and can’t tell us much of value.

That blue Corsi For bar looks very consistent because it is. The highest it gets is 51 per cent rounded off, and the lowest, this year, is right on 50. And yet we know that’s tank, wildcard, third place.  Mediocrity has many faces, and there’s a lot of ways to get to nearly dead even in shots.

The important thing to see in this graph is the dark purple Corsi Against. It is also nearly identical across the board, with all three periods rounded off coming in at 58 CA per 60 minutes. I had thought the Leafs had improved this year in shots against, and they have not. They have never changed since Mike Babcock’s first year as coach.  They are higher on the list when compared to the whole league now than last year, but that just means more of the other teams are performing poorly.

So far this year, the Corsi For for the Leafs is down a very small amount, enough to move the overall percentage down one point.

Last year, the Leafs were easy to figure out. Every quarter, including the last, which I never posted an analysis for, the answer was the same. Fluctuating goaltending and defensive execution sometimes covered it up, but the Leafs were allowing too many goals and too many shots against while producing near league-leading offensive pace.

This year, it looks like the offence has dried up, while the other problem hasn’t changed.  More context should tell us if that’s true or not, but it’s hard to believe the team is worse when they are in third place in the league.

More on Expected Goals and Scoring Chances

I have taken the opportunity of a new season and a seemingly stable supply of Expected Goals data to use that for context rather than Scoring Chances and High-Danger Scoring Chances as I’ve used in past years. The definitions that Natural Stat Trick uses for SCF and HDCF date back to War on Ice days, and have been used by many people for years. The divisions are based on shot location only, and are hard lines on the ice. So a shot judged by the official scorer to be a few inches over the line between the High-Danger and Medium-Danger area is suddenly seem to have significant difference from one inside it.  Expected goals eliminates that problem (which is called binning of the data) by weighting each shot individually.

You should think of Expected Goals information as a kind of grey average baseline to compare actual results to.  Actual goals scored come about because of skill, luck, random factors beyond imagining, and the actions of all players on the ice.  The reason we don’t predict future success based on past goal scoring is because it’s got too much stuff in it you can’t count on.  Shots have more skill and good systems in them, and less of the other stuff.

Expected Goals is what would happen if you replaced every player on the ice with a league-average version of a hockey player and then set them to work in a video game world with no random number generator used to spice things up.

Shots, Expectations and Real Goals

To get context on the way the Leafs are playing, we can look at the Expected Goals percentage and the actual Goals For percentage.

The real Goals For shows the progression from tanking to third place nicely.  And so does the Expected Goals.  Last year it was just a hair over 50 per cent, with the actual Goals For a hair over that.  This year the gaps are bigger. We can also see that the good CF percentage in the tank year was partly an illusion. That year’s xGF is below 50, and that means the quality of the execution by that team was poor, with the reality of the goal share showing that they were bad and likely very unlucky too.

For this season, the quality of the execution has been very, very good, and the results more so.  This is where things become difficult. When Goals For outstrip the Expected Goals, is that luck or player and team skill? The answer is nearly always that it’s both. It’s very rare for a team that is genuinely below average to sustain above average scoring for a long period of time. It happens, but you usually see a Corsi For well below 50 per cent as the clue that the team is riding a hot goalie or a lot of luck.

Goalies and Goals Against

You have to look at 21 games of Save Percentage information or Goals Against with a very skeptical eye. Last year we saw dramatic swings in performance at the goalmouth for the Leafs. The average of the whole season ended up fairly good, but at this time last year, we thought the world was ending.

This year, I’m showing you Fenwick Save Percentage and the Expected version of same. The Expected is an average, if it’s exceeded, that’s a mix of luck and a goalie with above average performance. If it’s not, that’s the opposite.

You can see where that tank really came from. You can also see that last year the overall goaltending results were a touch above expected. Please note the scale, these are very small differences.  Right now the goaltending is very slightly below expected given the types of shots faced.

Some context for Expected Fenwick Save Percentage numbers is really important here as most of us don’t have an idea in the back of our minds of what a “good” number looks like. The statistic describes what a league-average goalie would do when placed in a team’s net for every game. and the Leafs’ number is the third worst in the league, making the Leafs net the third hardest gig in the NHL this season—and tougher than last year. One team with a nearly identical result so far is the Pittsburgh Penguins, a point worth pondering.

I can’t look at any of this so far and conclude that the Leafs have improved defensively. They haven’t. They’re better than they were in November of last year, but all they’re doing is driving play as badly as ever, executing a little worse than last year’s average while getting good enough goaltending to paper over the hole.

Shots that are Lucky and Shots that are Good

If we glance back up at that big red Goals For bar that shows how the Leafs are in third place (they have no loser points, just straight wins), then given what we know, we have to assume that it’s shooting that is winning games.

The Leafs are second in the league in Goals For per 60 minutes, behind Tampa by the slightest hair. The Leafs are also eighth worst in the league in Goals Against per 60 minutes, in with a bunch of really bad teams and the Penguins.

Expected Fenwick Shooting Percentage is just like the save percentage version only taken from the offensive point of view. This is the league average shooter making the shots the Leafs made in grey, while the blue bars are what the Leafs have actually achieved in goals out of all unblocked shots.

Tank to wildcard to third place has come on the sticks of the skaters in the offensive zone. So far this year, the Leafs are the number one team in the NHL by their shot quality which has shot up massively since the tank year.  They’re also outperforming expectations with real goals by a very big margin.  The Leafs, not surprisingly, have the highest shooting percentage in the league. The gap by which they’re outdoing expectations is only the fifth highest, however.

So how much is luck and how much is skill? Is this all going to fizzle like a popped balloon?

You don’t have Auston Matthews, William Nylander, Nazem Kadri, Morgan Rielly, James van Riemsdyk, Patrick Marleau, Mitch Marner and Tyler Bozak on a team and get to claim their scoring is all luck.

You also don’t get to fly that high without expecting to come down some. The difficult question is, again, where is the line between luck and skill?

This is a graphical representation of how much and how successful the Leafs players have been shooting.  Farther to the right is how much a player shoots, and farther up is their shooting percentage.  This time, it’s Corsi Shooting Percentage, but that’s fine.  We are here to look at who is over-performing in where we imagine their skill to be. We have to imagine, because the Leafs have multiple shooters with very little career history to tell us for sure what’s normal for them.

Based on last year’s numbers for forwards and defenders in general, the defence is shooting in the normal range, and anyone higher than Hyman is above average for forwards.

Obviously Connor Brown is not going to keep floating up in the absurdly high range. We should expect Dominic Moore to drop a lot, Matthews to drop some, and some of the others up high to slide a small amount. There’s no schedule for this. It’s likely to happen, but the “when” of it is as up in the air as Brown’s numbers.

However, it is also very likely that Nylander and Marner will not stay down with the defencemen forever. Considering just how much Nylander shoots, a little regression on the percentage, some improvement in his shot locations and suddenly that’s a lot of goals that Connor Brown doesn’t need to be scoring to keep the overall team rate high.

Add to that that a lot of goals come on the power play—and the Leafs have an excellent one that we should expect to produce results in the future at a good rate—and it’s possible to think that the goals for will continue to outpace the goals against.

Conclusion: Is it Real and Is it Sustainable?

The Leafs outplayed their weaknesses last season with offensive pace, well enough to get a wildcard draw in the playoffs. Right now they have to outshoot those same weaknesses to keep winning, so is this third place stuff real?

Somewhat, is my answer, and yes, I would like syrup on that waffle. The pace of the offence had been much higher before it cooled off. It can pick up again, which helps make scoring easier, but that shot quality is very real.  The Leafs are good at offensive zone execution and at scoring goals, and some of their best at it are nowhere near their peak ability this season.

The Leafs are very bad at defence, and that’s really unlikely to change quickly.  When your top line features two guys who are nearly rookies and have spent their lives as offensive savants, you don’t expect them to be good at defence. And they aren’t, but for offensive savants, they’re better than you’ve a right to expect. Add to that the defence by committee, the secondary scoring line full of guys who think “glass cannon” is a compliment, and the Leafs are what they are: The best damn glass cannon anyone ever built.

Will this shatter in a million pieces?

You know, it totally can. It can also be held together by goaltending that’s good enough, some growth in defensive execution, which is a plausible ask from the Matthews line, and by even a modest upturn in offensive pace overall.

My prediction: The Leafs will keep on keeping on at about even on the shot differential, but the quality of the defence will improve enough to offset the inevitable drop in shooting success. Look, it’s just really good quality glass, okay?

Analysis