An Early Look At Leafs Player Usage

We’re 15 games into the season and we’re starting to get a clearer picture of what kind of team the Leafs are.  They sit in the final playoff spot in the Eastern Conference and have a record of 8-5-2.  They have a Corsi of 47.52%.  Guys like Phil Kessel and Leo Komarov have been awesome, while others like Stephane Robidas have not.  For this article, I wanted to take a closer look at one small part of this Leafs team: how Randy Carlyle and the coaching staff are utilizing the players.

THE CHART

Below you’ll find what is for the most part your standard usage chart.  On the x-axis we have Offensive Zone Start % and on the y-axis we have Corsi Competition %.  Most of you probably know how to interpret this chart already but for those who don’t here’s a simple rundown: the closer to the top of the chart you are means you face tougher opposition, and the farther to the left you are means you have fewer offensive zone starts.  So for example, the players in the top left of the chart are playing the toughest minutes while those in in the bottom right are playing the easiest.  One important note is that unlike most usage charts, both the color and size of the player circles represent TOI/G.  Usually the color represents Corsi, but I just wanted to focus (for now) on how the players are being used.  Players that play a lot will have a bigger circle and a deeper shade of blue, while players who don’t play a lot will have small red circles.  So, without further ado, here’s the table:

usage

THE PLAYERS

You can mostly tell which players play with which based on their proximity to one another.  For example, Tyler Bozak, Phil Kessel, and James van Riemsdyk are all closely cluttered to the center top of the chart.  To make it easier to tell who is playing with who though, via LeftWingLock.com, here’s a list of the Leafs most frequent lines and pairings (based on percentage of even-strength ice-time together):

James van Riemsdyk-Tyler Bozak-Phil Kessel 18.02%
Leo-Komarov-Mike Santorelli-David Clarkson 11.03%
Daniel Winnik-Joffrey Lupul-Nazem Kadri 8.98%
Richard Panik-Peter Holland-Brandon Kozun 2.64%

Dion Phaneuf-Cody Franson 21.58%
Morgan Rielly-Roman Polak 14.17%
Jake Gardiner-Stephane Robidas 11.32%

INTERPRETATION

  • We’ve seen a bit of a juggling of checking lines this season, and that is reflected in this chart.  It looks like the Leafs are kind of rolling a 1A and 1B of checking lines, with the first line facing tougher opposition but getting more favorable zone starts, while the (when the lineup is healthy) third line of Komarov-Santorelli-Clarkson is facing somewhat easier opposition while getting tougher zone starts.
  • I gotta say, this chart just makes me appreciate Roman Polak even more.  He’s looked solid for the most part this year and here you can see that he’s playing the toughest minutes on the team while also playing the most even strength minutes.  Stuart Percy, who is paired with Polak when he is in the lineup, is also playing very difficult minutes (pretty impressive since he’s a 21-year-old rookie).
  • The Leafs are doing something interesting with their fourth line.  Panik, Holland, and Matt Frattin are all facing easy opposition but getting really low offensive zone starts.  In other words, Carlyle is mostly using his fourth line against the other team’s third and fourth-liners, but he’s doing it in the defensive zone.  So he’s saving the offensive zone starts for his more offensively-inclined players, but he’s doing it in a way so that the fourth line doesn’t get burned.  I like that.
  • Generally speaking, most of the players are playing pretty similar minutes, not on one end of the extreme of competition or zone starts spectrum.  That goes back to the whole idea that’s been mentioned about the Leafs wanting to spread it out and not overburden their top guys.

IS IT WORKING?

Here’s the same chart, with the color of the circle now depicting Corsi Rel %:

usage2

Now here’s a comparison of the above chart (this season) with what it looked like last year:

usage4

There certainly are some improvements.  For one, the top pairing (Phaneuf and Franson this year, Phaneuf and Carl Gunnarsson last year) are doing a lot better with easier minutes.  You’ll also notice if you examine the numbers on each axis that Leafs players this season generally are playing easier minutes both in terms of competition and zone starts.  And while the Leafs team Corsi% isn’t as high as you’d like it to be, it is a full 4.70% better than last season.  How much of that is due to usage and how much of that is due to both a change in player personnel as well as a mild shift in team system though?  That’s hard to say.  Again, I didn’t want to so much criticize the coaching staff with this article as much as just highlight how the coaches have approached player usage so far this season.  It will be interesting to see how much this usage shifts over the course of the season as the team succeeds or struggles and as the lineup continues to change by way of performance, injury, and transaction.

Need help understanding fancy stats, check out this Analytics Cheat Sheet

  • Poluza

    This is some great analysis.

    I was very skeptical of the polak trade when it happened. But the fact that he’s handled his tough minutes so well has been an awesome surprise. Hopefully his good play keeps up!

  • BubbaLou

    You just can’t really notice things like Offensive starts, “Relative” corsi, and tough matchups until you see it side by side in chart form.

    Some other things to watch – and again big disclaimer: it’s 15 games. There’s like 4x more of these to play.

    -Is it just me or does this year’s team’s stats look a lot more even across the board? Maybe that’s a factor of removing players like smithson/gleason/fraser/smith that everything looks granular now.

    -Daniel Winnik both stats up as an important player, and quietly looks the part on ice. He’s the reliable two way winger/assist-o-matic, and is trending to a career year. Does that hold? (And would a tumble in november then have something to do with it if it doesn’t?)

    -If you follow +/-, Phaneuf is a +8 so far. Has he scored enough so far? No, he’s still tracking 40+ points but just 1 goal through 15. But has his line been lit up this year? Also no. That’s a good sign.

    -More on the top D: Last year, Cody Franson put up 33 points in a train wreck of a year. His performance was up/down/up/down/up/down all year, and we can all remember a game with a Franson screwup from last year. He’s been much more consistent this year (though he had a real goat game recently). What is the top D line going to look like months from now?

    -Jake Gardiner was great last year in generating offense and this year he has started out very average.

    -Leo Komarov is being used on ice to PK and to close out games for his Defence and his Corsi is slightly negative relative to the team. You can overlook this because so far Uncle Leo is awesome and a huge boost to play in our zone. From the Dangle Rangers video, DID YOU KNOW when Leo Komarov was last here, he did not play the last 17:51 of the Game 7 game? Seems almost criminal.

    And there you have it. We are getting 4.7% more offense than last year with a very similar lineup. It is both a huge boost from last year and still just elevates us to average.

    Why huge? It means more chances for us, closer games, and in turn takes some pressure off our goalies. Why average? 5th overall in Goals/game. 15th overall in GoalsAllowed/game. 14th in the standings… still league basement 27th in generating shots. The way we play, we need to convert a lot (but we are generating more of them now) and our goalies still have to steal games.

  • Poluza

    Your data makes me conclude that leaf fourth line is garbage. They are about on par with Colton Orr and Mclaren. Why would any analytics based coach play want the 4th line more then a couple minutes a night if they are such puck possession black holes?

    Using math, playing the fourth line more would simply drive down the leafs overall corsi. Wouldn’t it be funny if Carlyle 4th line usage was driven by some analytics like corsi or net scoring chance creation.

    • BubbaLou

      Well, for starters production is up. All Mclaren/Orr/Leivo/Ashton managed last season was 6 points – Half were Carter Ashton assists with.other.people. This year, Panik has 4 points already, and there’s a revolving door in the last winger spot chipping in 3 more points (because randy splits this line often as per above – and, really, we’re still going on the same 5 minutes a night Orr would get). So even if they all tore MCLs tomorrow, 15 games in you have accrued all the points last year’s 4th line produced in an 82 game span. They are unquestionably better at hockey than last year’s roster.

      One thing you have to consider, leaferfan, is the bottom row – zone starts. Generally, you give the benefit of the doubt to players bunched on the left side of the chart – those are the players who start face-offs in our end. That means there’s a good chance they start a shift by being outshot, so they get tanked.

      People I know call this the Grabo effect because it’s the most token case in Toronto of “mis-usage driving the stats” there is. Zone Starts have a big influence, no matter the team, on how many shots you will get as a player (and thus how many go in, and also how many goals get scored on them) so don’t write off people who aren’t given much offense just because they aren’t seeing much offense.

      More on that point, holy crap Daniel Winnik! Panik, winnik, santorelli – together are paid less than Bozak. Moneypuck is real, son!

      Finally, while there’s less stupid ice time usage (like McClement playing more minutes than Lupul or Kadri) this is still a Carlyle team and he wants 4th liners playing 4th liners – but only on defence. Besides an odd night here or there, like Panik on the 1st line, that’s all they run on.

  • BubbaLou

    I don’t follow the argument here. First, if we want to use similar zone starts then let’s focus on mclemment and smithson who had much harder zone starts in place of orr/mlaren. This year the 4th line generates more points then the 4th line last year but with a similar terrible possession and slightly easier zone starts.

    But if you argue for “better” offence, then that implies a “shot quality” argument on the team level (well the 4th line). Regardless, if this years fourth line is similarly poor at possession (on a relative basis) as mclemment and smithson then I don’t see why you want to give either of the 4th line (last year version with mcclement/smithson) who were buried with zone starts or this version) more TOI as both are struggling at possession.

    • BubbaLou

      Sure, let’s start there. I like to combine the statistical data with what I saw on the ice, the players’ stated roles by management and what they actually did. Don’t forget, also, that large portions of the season had the 4C between Orr, FML, or often times both at once. Players like McClement and Holland, for example, played on a modified shutdown line as a pseudo-3C and also the 4C fighter shepherd. It’s a shame there’s no TOI colour chart for last year here – I’m going off circle size (TOI) – but the colour comparison would help this argument because…

      Jerred Smithson is about as outlier as it gets for TOI. He was brought in as an emergency center when Kadri and Bozak were both injured in the same week, and to take draws like a David Steckel. Unlike Steckel, though, Smithson was not known for skating, or shooting, or even hockey – his job was literally just to win a draw and get to the bench as fast as possible. Look at his Ozone starts – 30%. In short, his data is skewed both in sample size (seconds of ice time), and usage (Faceoff win – get to the bench), so it’s almost impossible for him to “generate” offense. He also produced nothing points-wise despite all the sheltering in the world, but again how could anyone play like that?

      Now McClement, different story. He had tons of time on the 3rd line, saw all PK situations (and no PP), and served as the “responsible shutdown center” after Grabbo left. Clearly we see he’s the most “buried” in zone starts on the team that year and that really hurts. Despite the heavy defensive focus, he put up 10 poins -4G, 6A. IMO, he was pretty decent at this role. But like you said, the TOI doesn’t match the level of player. Is he Dave Bolland – no, but he played about as much. I actually liked what McClement brought in his own zone and thought he was a good roleplayer.

      You can ignore Smith, Ashton(who wasn’t that bad anyway), and Kozun – they’ve had only short stints and have been sent down since.

      So to answer your question – the role you play (or rather, the role you are used for) has a big effect on possession stats. Our 4th line is generally bad at possession because they’re mostly used on defense, and, well, they’re 4th line level players playing development-type roles. At least they’re buried competition-wise – unlike McClement they get a chance to play against their level of player.

  • May_Day

    Great article!

    In my mind, a good defensive defenseman is one that I don’t notice on the ice. Polak has been that this year, and it’s great to see it expressed on a graph!

    My one critic as a man of research, if you use the same scale for the axis across all graphs it makes it easier to compare 😛