15

K is for Killer Instinct

The Vancouver Canucks may have an orca on their logo, but the only thing they seem to share with a killer whale these days is that they spend most of their time under water.

At their heights from 2010 to 2012, the Canucks were ruthless when it came putting away there opponents. They liked to score early, and score often—never seeming to take their foot off the gas.

But in the fall to their current depths, the Canucks have lost that killer instinct. Not only do they rarely jump out to an early lead, even when they do, they aren’t able to hang on to it.

To get to the bottom of this, we’re going to have to take a deep dive…

Let’s start by taking a look at how teams across the league perform when they get a lead:

The first thing to notice is that there’s a pretty good relationship (surprise!) between being able to build on a lead and winning hockey games. The more you keep scoring after you go up a goal, the more points you will accumulate over the course the season. Based on a simple linear regression, the ability to build on a lead can explain about 31% of the NHL standings. The rest will be explained by other factors such as how well you score when tied or trailing, on the powerplay, your penalty killing, etc.

But our focus today is on how the Canucks play when leading, and the short answer is not good. In fact, only the Florida Panthers have a worse goal differential when leading by one goal.

This is a huge problem for a team that has such a hard time scoring at all.

It’s bad enough that when playing with the score tied, the Canucks have given up half a goal more than they’ve scored per 60 minutes of even strength time:

 

 

But no matter how much Travis Green claims to the contrary, when the Canucks do manage to get a lead, they fall back into a shell defensive shell and are much more likely to give up the next goal rather than to score another one themselves.

Here is how the Canucks perform relative to the rest of the league at scores ranging from trailing by two or more goals, up to leading by two or more:

The yellow bars represent the range of values for every team in the league at each score state, from lowest to highest, and the orange line is the league average with the error bar indicating one standard deviation to either side of it. So two thirds of the league, or 20 teams, should be covered by that error bar.

The good news is that the Canucks have been quite good at coming back from one goal deficits this year. They are just on the edge of one standard deviation above the league average, which puts them near the top five, and certainly top 10. They’ve also been pretty good at holding on to leads on those rare times they’ve managed to get up two or more goals. But it’s a struggle to get there. As stated earlier, they are near the bottom of the league when leading by one goal. Similarly, when they down by two or more, you can see that they essentially throw in the towel, which should be no surprise to anyone that has sat through those blowout losses as of late.

So what’s going on here?

Well, for one thing, we know that score effects are a thing. Every team in the league goes into a defensive shell to some extent when playing with a lead, and tries to pour it on when they’re down. Here’s what happens with shot attempts at different score states:

It doesn’t matter whether it’s a bad team (to the left) or a good team (to the right), everyone across the board has a better shot attempt differential when trailing than when leading. In fact, even the jumps from one score state to the next are fairly consistent at about 10 shot attempts per 60 minutes, give or take.

So let’s take another look at how the Canucks compare to the rest of the league:

Again, the team is better than average when down a goal, but extremely poor at all other score states, especially when leading by one, where they are more than on standard deviation below average.

But this is where things get interesting. Those two charts above were looking at the shot attempt differential (shot attempts for minus those against) per 60 minutes of play. If you split that differential up into it’s two components, a story suddenly starts to take shape:

When leading by one, the team is about average when it comes to preventing shot attempts against. Interestingly enough, the Canucks are also one of the best teams at limiting shot attempts against when they’re trailing by one. To me, this says they do quite well at limiting the counterattack when they are pressing.

However, when we look at the other end of the ice, we see where the real issue lies:

The Canucks are the worst team in the league at generating shot attempts when leading by a goal. While they do not give up more shots than average, they appear to be truly content to clear the zone and sit back rather than trying to generate a counterattack to build on the lead.

Given that the team performs so much better, both in terms of shot attempts and goal differential, when they are trailing by a goal compared to when leading by a goal, or even when the score is tied, let’s take a look at how Travis Green has deployed his players at those score states:

While Green might not be explicitly asking his players to sit on a lead, his deployment choices speak for themselves.

And so do the results we have seen so far this season.

 

 


RECENT GRAPHIC COMMENTS

  • defenceman factory

    Sure nice to read a Graphic Comments article about hockey and based on analytics.

    Very interesting to see just how poorly the team does with a one goal lead.

    I don’t think the deployment as % of 5v5 TOI is an indicator of Green deploying players to protect a lead. The better offensive players get 5% more ice time when down a goal vs up a goal. That seems normal and I expect every coach uses his top line a bit more when losing. In doing so the other lines get less of their minutes when down a goal. Why would a coach play Gaunce, Chaput and Dowd when down a goal?

    The Sutter and Dorsett numbers are at the top of the list but this is a pretty small sample size and for a short time it kinda worked.

    The conclusion here is more about the lack of depth and just how anemic the bottom of the line-up is rather than any critique of Green’s deployment. He simply can’t play 2 lines all game.

      • crofton

        I think the best coaches are able to assess their players’ strengths and weaknesses and devise and coach a system that considers that. And along with the rest of the coaching staff, perhaps a goaltender coach that doesn’t seemingly try to re-create goalies in his own image.

  • TheRealPB

    You can’t really pretend that this is a rigorous analysis if you make no mention of the context. Playing with the lead when you’re icing your full complement is a whole lot different than playing with the lead (or from behind) when you are down your top two centers, your top left winger, and your top pairing defensemen, as the Canucks have been for a good chunk of the season. The starkest difference I see in games like tonight is what Tanev brings to our d-corps and what Baertschi brings to the forward group in terms of both offensive creativity and greater defensive responsibility. It trickles down throughout both the forward and d group and you cannot talk about Green’s game plan without acknowledging the actual personnel. You’re discussing percentage deployment 5×5 TOI but include players who’ve missed half the season, are injury-retired, and have left the continent. TG’s deployments do not, in fact, “speak for themselves”; they speak to the actual personnel available to him.

    • Cageyvet

      Yeah, pretty “meh” statistics. Context is king, and I say this tongue in cheek, but maybe were really good at playing from behind because we get a lot of practice at it.
      It’s interesting to see the stats, and some of them intrigue while others just seem to be a numerical expression of common sense.
      I’d like to see some attempts at explaining how some statistical projections have failed so badly, and why, if they’re the outliers that you expect in most statistical models, we should place so much faith in these numbers.
      When I see more informed people than me debunk some of the stats, it’s to point to Pittsburgh winning cups with poor team Corsi, and similar oddities that require context to explain.
      Any hockey fan who watches the games knows that the stat line can be deceiving, and the attempt to remove randomness also excludes the human side of the game. Why does nobody try and chart momentum? I know that’s pretty damn tough, but any fan knows it plays a massive role in the outcome. Individual performances can look identical on the stat line but be completely different in their effect on the game.
      Take a goalie who faces the exact same number and quality of shots but lets in a soft goal (we know a little something about that). If it’s early, it sucks but you can recover. If it’s late, it flat out costs you the game. Everybody talks about a goalie’s ability to make a timely save, but there’s no attempt to track it that I’m aware of.
      Stats are great, keep them coming, but keep exploring better stats and not just refining possession and shot attempt metrics to the Nth degree. It seems like so much of the discussion revolves around regression to the mean and telling why a player will not sustain his good or bad performance instead of trying to identify those who buck the trend more often than not. That’s a skill that would get you a job with an NHL team, because identifying the Jamie Benns and Shea Webers is worth more than identifying McDavid or Doughty.

      • crofton

        I think trying to statistically quantify the human element is one of the largest failings of the analytics community. And it’s probably not possible, but like context, needs to be considered.

  • CanuckleheadOz

    Hmm, is my memory poor, or is this possibly peak Graphic Comments on this site?
    Interesting and seemingly without a flagrant agenda of castigating LinBennCoach?
    Hopefully this might be a sign of things to come or possibly even an indication of a more balanced and nuanced approach which has also been hinted at by JD recently. A minor editorial shift? Keep it up!

    • Bud Poile

      CA writers have had to change direction since the owner backed Benning in November on Sportsnet 650.Burke,MupPet Bugs and Jackson lost their battle:
      “We have a general manager,” he said of Jim Benning. “He is the expert. You have to go with the expert.”
      Aquilini said contract negotiations are ongoing with Benning and did not give details, “but as soon as we have something, you will be the first to know,” he told Walker and Rintoul. “I am pleased with how Jim has performed. I’m optimistic we’re going to come to an agreement with him.”
      http://www.sportsnet.ca/hockey/nhl/canucks-owner-talks-sedins-jim-benning-possible-nba-return/