Friend of the blog and former MHH regular horbayj asked on Twitter today whether the Avalanche have been leaning less on Semyon Varlamov recently. After all, the team game looks better, and results are following, and we know from experience that relying on Varly to bail us out of every game isn't a sustainable strategy.
The quick answer is yes, they are leaning on their goaltending less. The rest of this article attempts to understand why. Literally all numbers are taken from War on Ice, without whom crunching a bunch of season long numbers like this in 2 hours could never be possible. <3 u guise.
That's the Avalanche's 10-game rolling average shot differential. It's not the first time Colorado have outshot their opponents over a 10-game stretch, but it is the second, which makes it significant.
We have noticed in this space before that the first bump in shooting came right at the exact same time that Brad Stuart went down with an injury. What I think nobody expected is the exact same bump showed up again when Erik Johnson went down.
(NOTE: I actually put the Stuart injury in the wrong spot here; it should be one trough to the left. Above "stuart," not "injury." Also this is score-adjusted.)
So what the hell, man? How can it possibly be that the team possession game improves this dramatically when it loses its worst possession impact player in Stuart (-103.57 dCorsi Impact), but does the EXACT SAME when it loses its 4th-best in Johnson (+53.26 dCorsi Impact)?
I'm focusing on the defensemen here for two reasons. We're talking about changes to the makeup of the blueline. Second, forwards would literally add twice again as many players to the analysis. This is quite enough to look at as-is.
Usage as we understand it today comes down to two major factors. Who plays how much, and where they get deployed. Competition is often overstated in the context of entire seasons, and can be implied in raw TOI honestly. Guys who play a lot see a lot of other guys who play a lot, and coaches by and large (although fans will always scream about exceptions) play the best players the most.
So let's start there, with the rolling possession graph above plopped on top of a chart showing each defenseman's share of even strength ice time. I tried to organize this by current pairings but they have definitely shifted over time.
Let's note first that the TOI chart is game-by-game, while the possession number is a 10-game average, so comparing any one data point to another is a bad idea. We're looking for trends here, and a few jump out:
- For the most part, TOI has been very stable on the back end. Apart from players entering and leaving the lineup, the changes here are, past about game 8, just game to game noise. A notable exception is Tyson Barrie taking big minutes away from Nick Holden around games 25-37.
- When the team possession heads upward, the defenseman being added to the mix has been Zach Redmond both times.
- Possession was already starting its tail spin long before the return of Brad Stuart. The big change around game 25 appears to be the ice time of Nick Holden.
We'll return to Zach Redmond momentarily, but first, the unfortunately much uglier chart of defensemen's defensive zone start shares, again with CF% superimposed.
There is some massive game-to-game noise among players who don't take a lot of the D-zone starts. It kind of helps to look at this one for trends among who is taking the most.
- Reliably, dumpster starts have been given to poor Jan Hejda, who has at times started as much as 40% of the team's defensive zone draws, but not so much lately.
- With the exception of one game here or there, you'll notice that Stuart and Holden have seen their shares of D-zone starts increase lately, as opposed to the beginning of the season. We see them taking about half of the faceoffs in front of Varlamov now.
- Tyson Barrie and Nate Guenin's zone usage is so sheltered it's disgusting lol.
- Erik Johnson taking more than 25% of the defensive draws himself seems to result in extremes. Either the team possession is great in those stretches or it's awful, with very little in between.
So taking these two visualizations together, and ignoring the stellar play of Ryan O'Reilly's line for our purposes here, let's find some things to take away from how the defensemen are used without a big piece and how the team can use that knowledge to improve when they return.
First, Zach Redmond is quite good offensively and needs to remain in the lineup. How he is used matters. Redmond generally gives up about as many shot attempts as his situation would expect, maybe very slightly more (dCAImpact = 3.53), but he is a generator of offense (dCFImpact = 19.12, 5th on the team in 10-20 fewer games than everybody else). He has been pretty sheltered so far, so the team might benefit from his usage becoming maybe more neutral.
Let's stop using Hejda in the defensive zone all the time. Johnson can definitely suppress shots (the only better defenseman in terms of dCA Impact is Barrie), but Hejda doesn't really anymore. Hejda's shot suppression vs. expected is actually worse than noted whipping boy Nate Guenin's at this point. Sometimes, it might be useful to swap Hejda out when the opponents' best are coming over the boards for a faceoff in front of Varlamov.
Sheltering Barrie this hard might not be helping. He generally never has more than 15% of the team's defensive zone starts, but the only Avalanche skater who suppresses shots vs. expected better this year is of all people Nathan MacKinnon. (MacK has had a quietly excellent time possessing the puck this season relative to his teammates.) You'll need to be careful about deploying Barrie in his own zone against smooshy players like Byfuglien or the Backes line, but never putting him there at all restricts the team's ability to leave its own zone.
It's going to be about making do with the players we have, and the sad facts are that this blue line doesn't possess the puck very well. However, this is about making do with what we've got. If the team shows it can make do to this extent without Erik Johnson, maybe it's not so far off from being at least okay when he returns, with some tweaking.