clock menu more-arrow no yes mobile

Filed under:

By the Numbers: A Tale of Two Bloos Games

The Avalanche stomped the Bloos before Christmas, and got casually run over by them after it. What changed? WARNING: Charts, Words

SPOILER: i talk about redmond and o'reilly in this
SPOILER: i talk about redmond and o'reilly in this
Chris Humphreys-USA TODAY Sports

Last week, the St. Louis Bloos came to town for a game at the Pepsi Center, only to get run back out of it again. The Avalanche won 5-0, dominant from the word go, outshooting the Bloos in every period. On Monday, Colorado visited St. Louis, and while they may have only fallen 3-0, the game was never really close after the first period. So I thought it might be illustrative (and help kill some time before games) to roll up the sleeves and dig into the numbers to look for differences.

O'Reilly sure looks like a dwarf in that picture doesn't he?

I like to start by looking at game flow, as you know. So here's running Corsi differential (even strength), first for the win:

col 5 stl 0

This is a standard Big Win game. Take over early, push repeatedly, score some goals, and at the end it all slows down and score effects do what they do. Colorado peaked +16, finished the 2nd period +14, finished +5 after an inconsequential third period, and never trailed the possession game. Against St. Louis!

On the other hand, the flow for the loss. At even strength, there were 13 total SOG in the first, 13 total SOG in the second, and 14 total SOG in the third:

col 0 stl 3
Things got out of hand very quickly in the second here. Before, the game was slightly in St. Louis's favor but mostly even, and after, the game was boring as hell because that's how Hitchcock teams protect leads. Colorado had 8 SOG in the first period, and 8 SOG in the rest of the game.

Comparing the sheer context of these games gets us nowhere. There wasn't much to indicate either game was a schedule win. The home team has about a 55% win rate, and in neither game was either team on a back to back. In the Avalanche win, St. Louis was actually more likely to win than a road team overall due to having an extra day of rest. [1]

So my next stop will be comparing some individual numbers. [2] Namely I want to look at possession, usage, and matchups to see how much we can explain. I have removed data for Dennis Everberg and Nick Holden, because neither played in both games. Let's start with who saw more of the ice. This chart is the difference in even strength ice time, extrapolated over what would be 60 minutes. I made that adjustment because there was an over 10 minute difference in the amount of 5v5 hockey.


Three things jump out immediately. O'Reilly's line with Landeskog and Briere saw a lot more of the even-strength ice in the 0-3 loss. Duchene in particular saw a lot more of it in the 5-0 win. Erik Johnson and Jan Hejda also saw less of the even-strength ice in the loss, which I thought was sort of weird. I would expect that in a game where they're also spending time on special teams, but not this one, which was practically all 5 on 5.

So let's compare player performance in terms of possession from game to game. First the fowards, then the defensemen, because this is quite a bit to look at all together.

forwards corsi diff

(Sorry this is tiny. I honestly can't figure out why you can't click to embiggen anymore. Hold CTRL and scroll to zoom.)

The biggest drop comes from Ryan O'Reilly, followed by the line of Alex Tanguay, John Mitchell, and Jerome Iginla. O'Reilly's linemates also fell, but not to the same extent he did. So not only was his per-minute influence much, much worse, he was given a significantly higher proportion of the 5 on 5 ice to work said, erm, magic. That's probably because they spent plenty of time killing penalties (to either team hahahaha sad) in the win, at least to some degree.

Usually when you see a big drop like that the first thing you want to check is usage, because nothing drives possession like the part of the ice you start in. Before we get to that, let's check in with the defense.

defensemen corsi diff

That's right! In a trashfire of a loss, Erik Johnson managed to produce better than he did in the roflstomp victory. What? Jan Hejda's rates remained about the same, if lower-event overall.

Zach Redmond is your worst dropoff offender here, in case people are wondering why he's scratched tonight. Following him are Brad Stuart and Nate Guenin. Redmond aside, these are not much different in the Against category - it's a dramatic downtick across the board for defenseman in the For category.

So let's check zone usage. Relative to the team (because the team had very few offensive zone starts in the loss), did anyone see significantly different zone deployment? (Being on the axis of this chart would mean you had exactly the same ratio of offensive to defensive zone starts in both games.)

ozone diff

O'Reilly and his linemates were deployed about the same. That's the difference of a shift or so from the overall ratio. Cody McLeod appears to be much more heavily sheltered in the loss, but I'd argue there are some sample size shenanigans going on there. The main differences in usage were John Mitchell seeing more starts in the offensive end, and Duchene's line getting buried a little bit.

Remember back in the beginning of the year, before the switch to zone, when Matt Duchene was getting all the defensive zone starts and getting killed? That actually didn't happen much in this instance. His line generated fewer events for, but did much better in terms of events against. Duchene in particular improved from 80.8 CA/60 to 46.8 CA/60 in the loss with much more defensive usage.

So what's the difference? Let's look at matchups. [3]

In the win the Avs' top lines opponents are kind of a hodge-podge. Looking in particular at O'Reilly, here's his most common opponents in the win:

  • Alex Steen (7:02)
  • Kevin Shattenkirk (6:59)
  • Alex Pietrangelo (6:44)
  • Paul Stastny (6:05)
  • 4 others with 4:45 or more TOI against O'Reilly

And in the loss:

  • Alex Steen (11:47)
  • TJ Oshie (11:34)
  • David Backes (9:22)
  • Alex Pietrangelo (8:57)
  • Jay Bouwmeester (8:24)
  • no others with more than 4:30 against O'Reilly

That, you guys, is line matching.

In the victory, Duchene saw 6 minutes or more of Chris Butler, Dmitrij Jaskin, Kevin Shattenkirk, Paul Stastny, and Patrik Berglund, with 5 minutes or more of 3 others. In the loss?

  • Kevin Shattenkirk (8:58)
  • Vladimir Tarasenko (8:11)
  • Jori Lehtera (7:26)
  • Chris Butler (6:59)
  • 5 minutes or more of one other (Patrik Berglund)

That, you guys, is line matching. By the way, the forwards who took the biggest dip as a unit? The Mitchell line? Yeah, in the loss they saw 10 minutes of Paul Stastny's line. The Curse of ex-Avs is goddamn insidious.

Hitchcock's line matching plan here is pretty plain. He has two strong possession lines, the stronger of which (Backes) gets to neutralize the Avs' stronger possession players (Landeskog, O'Reilly). The other (Stastny) takes the Avs' secondary scoring (Iginla, Tanguay) out of the game entirely.

Then what does he do with Duchene and MacKinnon?  Play his own risky line and roll the dice on the percentages. You know Duchene won't crush you in possession. With two other lines shut down and the fourth inconsequential, you can totally make that gamble and win, especially knowing your team is better systemically in the neutral zone. On this night their 5 shot attempts did not result in a goal. On another it may result in 5, but that's rare.

And that's my narrative on how the Avalanche got crushed by an opponent they pooped on a week earlier. St. Louis's coach did a better job matching lines up in St. Louis, and his plan worked out very well for him.

[1] Schedule effects data from Puck Prediction

[2] Individual data all taken from War on Ice. Win || Loss

[3] TOI with certain opponents can be found on Natural Stat Trick. Win || Loss