clock menu more-arrow no yes

Filed under:

Slicing open the Maple Leafs with Occam’s Razor

New, comments

Looking for the simplest possible explanation for the results so far this season.

NHL: OCT 25 Sharks at Maple Leafs Photo by Julian Avram/Icon Sportswire via Getty Images

No one is very happy with the way the Toronto Maple Leafs have played, not recently and not on balance for the season. Granted, the season is only 13 games over four weeks, so most of it has been very recent. But things don’t look right or feel right or add up to good in the standings.

Just the Facts

Seems isn’t the same as is, so what is the actual state of things?

The Leafs are 12th in the league, seventh in the Eastern Conference, and third in the Atlantic. As everyone predicted, the Buffalo Sabres are the number one team in the NHL right now. (Now is Sunday, so by Monday, this will be different.)

The Leafs have an ROW of six, which puts them 11th with a goal differential of +1, which puts them 17th.

At the moment, there are two really bad teams in the Atlantic: Detroit and Ottawa, and every other team is bunched up between 12 points and 19, making this a genuine six-team contest. The same is generally true in the Metropolitan Division, with only two very poor teams so far. Only six teams in the east have negative goal differentials.

The Leafs are seventh in the NHL in five-on-five Score and Venue Adjusted Corsi For percentage with 53%. The Flyers at the top have 56% and the Ducks at 15th are 50%. The Fenwick For % (all unblocked shots) is a little lower, with the Leafs at 11th in the NHL at 51%. I won’t be discussing full season Expected Goals until the data is fixed.

In Corsi For per 60 minutes, the Leafs lead the NHL with 62. In shots against they are 18th best at 55. That is the biggest change in any Leafs results in this season over last and marks a significant improvement in the one area of play they have been weakest at for years.

The Leafs have the seventh lowest number of minutes spent on the PP per game played. The Islanders should complain to the league because their 2:15 is absurd. The Leafs at 5:55 are only half a minute away from league average. They are 17th worst for limiting time on the PK (although, in one sense, more time can be better). Their 5:28 is within a whisker of league average.

I refuse to discuss power play or penalty kill success percentage. If you want results, then Goals For and Goals Against per 60 minutes is acceptable. On the power play, the Leafs are 16th with 7.5. On the penalty kill, they are 15th best with 6.75. In terms of Corsi on special teams, the Leafs are league average in Corsi For per 60 and slightly better on the penalty kill at Corsi Against per 60. Their special teams mirror their improvement overall in shots against, but showing a much different shots for rate than at five-on-five.

Note: throughout this article, shots, when used casually, means all shots, or Corsi. Shots on Goal will be identified as such.

Occam

The idea behind Occam’s Razor is that when you’ve got competing theories that lead to the same results, you should pick the one with the fewest assumptions. In other words, if you have to make up a complex narrative to get you from the cause you suspect to the results as they actually are, and that narrative includes a lot of unlikely, unproven or untestable assumptions, you should discard it in favour of the simplest answer.

The Leafs have just been unlucky

There’s a few different kinds of luck, and it’s not possible to even discern some of them. But here’s a few obvious ones:

The percentages: PDO (an acronym that means nothing) is the Save Percentage and the Shooting Percentage on Shots on Goal expressed as a decimal and added together. Over a large amount of time, this number will be 1 for all teams, seasons and the league as a whole because for every shot that’s a goal in one number, it’s a goal against on the other. The Leafs are at .987 at five-on-five and 1 in all-situations.

The Leafs are shooting the puck more than any other team, and yet their shooting percentage based on Shots on Goal is 8.67% at five-on-five which is 12th and just a touch over league average. On the power play is it 14.29%, which is 14th.

If you assume that the Leafs should shoot better than average overall, and definitely should on the power play, they are below their normal results and should regress up in time.

As for the goaltending, which is only two players and is therefore more skill-based, the results are not good. Michael Hutchinson is well below league average in results, and Frederik Andersen is better, but still in the bottom 20% of the goalies who have started at least three games. When you look at all-situations, both goalies look a little better.

Unless you assume Andersen will stay this bad (and there is every reason to think he won’t), then the Leafs have been unlucky in their goaltending so far. The Leafs have faced 426 Shots on Goal against so far this season (the highest in the NHL) and if they had league average goaltending, they’d have 39 Goals Against instead of a league-worst 44.

Verdict: The Leafs have been a little unlucky in shooting and a lot in goaltending at five-on-five. The PK goaltending is okay. This isn’t the whole answer, but it must be considered a meaningful part of the story, particularly the power play shooting and the five-on-five goaltending.

The schedule: While playing back-to-backs isn’t ideal, the parts of the scheduling that can randomly affect results in a few games are an imbalance in road vs home games or an imbalance in strength of competition. Home ice advantage matters more than most other things that affect any “on paper” comparison of two teams.

I used a bit of a cheat to calculate this, because it’s really hard to do in a meaningful way separated out from roster issues like the dreaded goalie question and injuries. I took the opening probability of the Leafs winning from Moneypuck on each game and averaged it. The Leafs have eight home games and five road so far and the average probability the Leafs would win was 49%. That means that given the strength of the opponents, the schedule and the roster, the Leafs “should be” at about a 50% points percentage in a world without three-point games. Their points percentage in the real world is 53.8%. But that ROW of six now seems a lot less bad than it did a minute ago.

Verdict: there’s been a few tough teams visit (including the Stanley Cup Champions and the runners up) but overall, the extra home games have almost made up for that. Expecting the Leafs to be running away up the standings chart is not fair, however. The schedule has been a small part of the story.

Injuries: The roster problems are largely folded into the schedule analysis by lowering the Leafs probability of winning. With a full roster from game one through 13, that 49% would likely be at or over 50. John Tavares missing games is having a huge impact recently, and the lack of Zach Hyman and Travis Dermott is having a smaller effect overall.

Verdict: getting John Tavares back is necessary for success, but his injury isn’t the biggest part of the story so far.

The Leafs have played badly

After looking at luck, the next most obvious and simple thing to examine is team and player performance. Most of that is described well by the facts listed above.

Driving play: The team shot share is excellent, and since it’s aligned with our understanding of the player strengths and weaknesses: fantastic shots for and poor, but not horrible shots against, we can expect that to be sustainable and not some random blip. For individual players, the results are also positive, with the good players who can score spending lots of time shooting, and the players who are worst at defending spending less time trying. No one stands out as being particularly bad, although the rumours of the fourth line’s defensive prowess are overstated.

Verdict: this is the best the Leafs have ever performed against a string of tough competitors. The poor results are in spite of this, and are therefore, worse than they look in some ways.

Defence: It’s very tempting to blame the defensive execution of a team for a run of bad goaltending, and that’s fair in a game or two, and it’s still a little bit fair in 13 games. But the Leafs have, over the games we have reliable data (six), been a little better at Expected Goals relative to the other teams than they have simple Corsi Against. Unless they were horrible in the first eight games, defence isn’t any more of a problem than it simply will be because the team is not very talented at that side of the game. Expected Goals doesn’t account for incidental stupidity, but that has to be seen as part of the random variance of results vs expectations (big E or small e).

Verdict: when you see a goal against, you see specific players to blame, but overall, the Leafs have been better than we have much right to expect defensively. This isn’t the story one way or the other.

Offence: As stated above, no one shoots more than the Leafs at five-on-five. So if there’s an offence problem it’s either shot quality or it’s the power play. It seems to be both and some bad luck as well, judging by the results.

Given the shooting percentages (based on Shots on Goal) that are only average, the Leafs are not riding a wave of shooting luck as a team. Some individuals are, which is normal over a small sample of games. The lucky ones who aren’t fourth liners playing very low minutes are: Alexander Kerfoot, John Tavares, Auston Matthews and Ilya Mikheyev. Tyson Barrie and Mitch Marner are snake bit, Andreas Johnsson is going to score a lot more than he is right now, and Morgan Rielly is back to scoring like a defenceman, not a forward.

Over the games for which we have reliable data, a finger can be pointed at Auston Matthews and John Tavares for not bringing quality shooting. Mikheyev and Nylander are leading the team in individual quality shots, but I think we can all admit they aren’t the best shooters on the team.

Examining the power play in depth is beyond the scope of this article, but the shooting percentages and the shot rates say there is a problem there. Matthews and Marner have been excellent, and no one else has been anything more than adequate.

Verdict: The offence is where the heart of the player performance on the Leafs has to lie, and it’s been substandard on the power play in all ways, and in shot quality at five-on-five while having an underlying pace that will, in time, provide a nice cushion of goals to tide the team over the rough patches.

Who done it?

Occam says the problem is that in a set of games with a difficult, but not overwhelming schedule, the Leafs have been hamstrung by bad and/or unlucky goaltending, and haven’t added to their exceptionally good five-on-five shooting with a good power play. Their shooting luck is not terribly bad or good, but it should come up.

Why won’t you testicularly challenged dudes tell MLSE what the real problem is (to misquote a bozo on Twitter)

Where is coaching at play?

This is the hypothesis as I understand it that says Mike Babcock has personally destroyed the team: He has made the players play a defensively-focused, grinder-style game and they’ve lost the ability to generate creative and fun offence because of him and his systems regardless of who the assistant coaches are or which forwards are on the ice.

This is the sort of thing Occam made his razor for. In order to buy into this, you have to create a scenario where despite the excellent CF% and a league-leading CF/60, the Leafs offence is so bad on quality as to overwhelm that into meaninglessness, and not only that, it’s because of the mental state of the players who are trying to play in wrong ways focused on defence.

This is, to be kind, a highly unlikely set of assumptions. There are some worrying results around shot quality, but they are not major concerns at five-on-five or caused by defensive focus. They are separate. There is evidence that at the player level, skill in reducing shots against and skill at producing shots for are not linked. The system a team plays (stretch passes, dump ins, carry ins, etc. etc.) either results in a good shot differential or it doesn’t and when it does, stop worrying.

Where I look to the coaching is around the power play, and the offensive systems as separate from defensive play altogether. It’s what’s happening inside the offensive zone that’s bothersome in a mild way.

If you look into how an Expected Goals model is made, where thousands of shots from years of data are analyzed and we gain an understanding of how you maximize your quality of shots, we can understand the following:

  • volume of shots is the most important thing by a large margin
  • shot distance is the next most important
  • rebound shots and shots on plays that force goalie movement are next
  • everything else is minor and not worth fussing over

The Leafs’ good players who have the skills to score are doing the first thing really well, the second thing is happening in a more league-average or a little above average way, and the third thing is occurring a lot less than they are capable of doing. The power play starts out with lower shot volume than we should expect and contains the same issues around quality.

This is where coaching comes into play and where Babcock and Paul McFarland have to look at who they have playing, how they’re playing and try to get the best out of them. Some particular attention should be payed to Auston Matthews decline in five-on-five effectiveness from ultra-elite to merely extremely good. There’s always going to be room to improve the defensive execution of a group of players whose main skills are not defence, so that needs to happen, but the goals against is a goalie issue overwhelmingly.

It’s been said that the Leafs aren’t trying, and I can’t see how that makes sense. No team that is leading the NHL in the rate of shots for is being lazy, indifferent or playing badly on purpose. If you’ve created a mental model of a bunch of Martin Marincin-esque barely inside the blueline shots to explain away the Corsi results, that’s false. They also aren’t only getting good Corsi because they’re trailing or against weak teams.

All of those ideas are sliced away by Occam’s razor. What’s left sounds a disturbing amount like the way the team played last year in the late winter when the goaltending got iffy, and there were injuries to key players. (The Corsi % was also bad then, however.)

Here’s my over-arching hypothesis: The Leafs need all of their top skaters and their starting goalie to be in the game and performing at their true ability most of the time or they are only a good team, not a great team. All the rest is irrelevant noise. Until that can happen, the best thing the Leafs could do with their sudden time off is have a long hard look at their offensive set plays and the power play design.

Then all they need is a run of games where the team has more than a 50-50 chance of winning half of them.