Winter Olympics 2014: The real reason P.K. Subban should be playing

Bruce Bennett

Why the reigning Norris Trophy Winner may not be the defenseman the world thinks he is, and why a shot-based analysis doesn't give us the whole picture.

There's been a lot of space on this site devoted to proving the value of P.K. Subban over the past few years, so I don't want to take too much more. But I think there's a clarification that needs to be made, both on the part of the statheads and on the part of the old-schoolers, because at this point wheels are just spinning aimlessly. Mike Babcock yesterday came out and said that whether it's Subban or Dan Hamhuis in the lineup the rest of the way, they're doubtful to see much ice. This (like anything else) led to a new series of articles and tweets about Subban and how he's so obviously one of the six best defensemen on Canada. In particular, this article over at Canucks Army was well-received in the analytics community - and it should be, it's a fine article - and I'm sure elsewhere it was dismissed with a "watch the games" wave of the hand.

Getty Images

One of the things I like to do as someone who grew up a hockey purist but have now adopted analytical viewpoints, is to rein both sides in a little to find the truth between the two extremes, and that's what I'll try to do here. So let me first summarize what I believe to be the most common versions of the arguments in favor and opposed to P.K. Subban being in the lineup (or in Canada's top 6), as I've seen them thus far.

Argument for Subban

P.K. is undeniably one of the best offensive talents in the league, but his shot prevention has actually been highly underrated (and in fact elite) throughout most of his young career. He doesn't have to do the conventional defensive duties like shot-blocking and hitting as much as most defensemen because of the time he spends in the offensive zone. In the end, his shot/fenwick/corsi differential is at the top of the NHL, and those measures tend to correlate highly with scoring chances and thus goals. In other words, Subban is one of the top defensemen in the NHL. He has also been very good on the penalty kill (when his coach has used him there). He should be playing.

Argument against Subban

You can't deny that Subban is a force offensively or on the power play, but you can contest the defensive aspect of this. Subban may not allow as many shots as others but he makes a lot of mistakes that lead to goals against. He's often out of position and doesn't make the best decisions with the puck. Maybe more importantly, he's a right-handed shot in a country with an excess of those on D. He's clearly not better than Drew Doughty and Shea Weber, and Alex Pietrangelo has the chemistry factor with Jay Bouwmeester and the experience and success on the penalty kill. Ultimately, it is difficult for a guy who's rarely played his off-side in his career (especially one who's already prone to mistakes) to do so in a one-game elimination tournament, and it makes more sense to have Hamhuis in the lineup who can take the odd shift from Marc-Edouard Vlasic or Jay Bouwmeester on the left side.

I may be missing a point or two, but those are the basic two arguments. And the thing is, I don't think either side is actually wrong. That doesn't mean that there aren't people who are wrong. I've seen a lot of "Subban is a defensive liability" and "he's not a top defensemen" and what-not, and those really aren't defensible statements. But this is not a situation where Subban is being compared to his Canadiens teammates or the league average. This is the best of the best. I think the righty-lefty point is a fair one (although I was a goalie when I played hockey so I don't know for sure how big of an issue that is), so the odds are stacked against him to begin with.

Using goal rates over shot rates in Subban analysis

I think we're at a point in hockey analysis where you can't deny the predictive ability of possession stats on broad basis, but I also think we've reached a place where ignoring shot quality, especially when it comes to defensive proficiency, is unwise. Chris Boyle has done a lot of great work to show that teams can impact shot quality against, and that starts with individual players. There's still much work to be done in this area, but when you look at an article like this, the conclusion needs to be "shot quality IS important, we just need to find a way to prove it".

Hockey Analysis had an interesting piece last year that found that goal rates actually correlate better with future goal rates when the sample size is multiple seasons. In theory this makes sense. Once you have a big enough sample size, the variation in shooting percentage is likely to diminish, and you get a better idea of a player's true shooting percentage and thus their true goal rates.

So rather than reference Subban's incredible shot rates, when you know that the response is going to be about shot quality, why don't we take a look at goal rates over a several-season sample, and then figure out whether Subban really has defensive issues.

Here we're measuring his goals for/20 minutes of ice time, goals against/20 minutes of ice time, percentage of total goals while he's on the ice that his team scores, and we're looking at the appropriate stats for 5v5, 5v4, and 4v5. These statistics are measured from 2010 to 2013 (I don't think Subban's reputation is based on this year alone so that shouldn't matter) and the threshold for NHL defensemen is 2000 minutes played for 5v5, 300 minutes for 5v4 and 4v5.


Subban NHL Rank Team Canada Rank
5v5 GF20 0.859 27th/145 3rd (2nd among RD)
5v5 GA20 0.650 19th/145 1st
5v5 GF% 0.569 11th/145 2nd (1st among RD)
5v4 GF20 1.951 42nd/75 6th (3rd among RD)
4v5 GA20 1.357 5th/100 1st

These numbers truly back up what Bruce Peter wrote about in early 2013, when the Subban contract negotiations were going on. What's important is that a) these aren't shot-based metrics, so the debate over shot quality doesn't apply, b) this is over a three-year, 2000+ minute sample. I'll also point out that his shooting percentage on the power play is not abnormally low, nor is his team's save percentage on the penalty kill abnormally high. This is a credit to Subban's defensive talent, and maybe even an indication that his offensive production is overblown (although the talent is clearly there).

There's no doubt it's largely a result of his strong possession play, leading to very little defensive zone time, but whichever way you look at it, Subban kept the puck out of his net more than any Team Canada defensemen the last three years. His good offense leads to good defense, but his penalty kill numbers show that he is also elite in his own end. He can play positionally, and he's good enough along the boards that he wins the puck back and rarely misses a clear.

There's no doubt that shot-based metrics have a place in analysis, especially over small samples, but if the debate is over whether Subban makes too many mistakes, then the best place to look is the scoreboard. These numbers show that however he's doing it - my guess would be incredible skating, an uncanny ability to win puck battles, and great stickchecking - he's making forwards look bad when they're on the ice against him. And that's why he should be playing.

X
Log In Sign Up

forgot?
Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

Join Eyes On The Prize

You must be a member of Eyes On The Prize to participate.

We have our own Community Guidelines at Eyes On The Prize. You should read them.

Join Eyes On The Prize

You must be a member of Eyes On The Prize to participate.

We have our own Community Guidelines at Eyes On The Prize. You should read them.

Spinner

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_9355_tracker