Evaluating NYR Defensemen on the PK

October 21st, 2014 at 6:00 am by Derek

Most of the time when we talk about hockey analytics, we speak only about 5v5 scenarios. This makes sense, because about three-quarters of the NHL hockey is played at full strength. However, special teams play are obviously important to winning hockey games. Though it is possible to hoist the Stanley Cup with deficiencies in your power play and penalty killing units (remember the Bruins 11.4% PP during their 2011 Cup run?), it sure makes winning a hell of a lot easier if one or both of them can dominate.

Let’s take a look at defensemen on the penalty kill. Last week, Bill talked about Tanner Glass’s effectiveness on the PK using Fenwick against per 20 minutes. I took a similar route for defensemen. I collected a bunch of stats from Puckalytics for defensemen who played at least 300 minutes at 4v5 between 2007-08 and 2013-14 and ran a few regression analyses:

PK_D_correlation_summary
For defensemen on a 4v5 penalty kill, Corsi against rate has a slightly better relationship to goals against rate than Fenwick and shot against rate. Intuitively, this makes sense: the more shot attempts that a team on the power play gets, the more goals they’re going to score. For a defenseman who is killing penalties, the name of the game is suppressing shots by forcing turnovers and sending the puck 200 feet down the ice.

However, raw Corsi against rate isn’t going to be wholly informative to us when looking at defensemen. Whatever system his coaches employ, as well as his quality of linemates, will have a big influence on a player’s raw rate numbers. Therefore, looking at relative stats will give us better insight into how well a defenseman is performing on the PK. A useful tool for this kind of analysis is CA60RelTM — a comparison of a players Corsi against rate to his linemates’ when they aren’t on the ice with him (more on RelTM stats here). Using CA60RelTM, we can get a decent idea of how well a defenseman suppresses shot attempts compared to his linemates when he’s on the bench.

Now that we’ve got the basis of our analysis set up, let’s see how the current crop of Rangers defensemen has played on the PK in the last four seasons. I set a minimum of 200 total 4v5 minutes; this is still a pretty small sample size, but PK minutes are inherently limited. Note that a negative CA60RelTM is good — you want to have fewer Corsi against per hour than your teammates when you’re on the ice. I’ve also included each player’s rank among all 160 NHL defensemen who fit my constraint.NYR_currentD4v5
First observation: good lord, Dan Girardi. Not only is he the worst defenseman on the Rangers by a country mile, he’s also one spot away from dead last in the entire NHL. The only player ranking below him, Brent Sopel, hasn’t played in North America since 2010-11. This isn’t a small sample size issue, either — Girardi has played nearly 854 minutes at 4v5 since 2010-11, fifth-most in the league. There’s no question about it — he is absolutely atrocious at preventing shot attempts on the PK.

Otherwise, the stats aren’t too surprising. Kevin Klein doesn’t look good here, but most of his PK minutes in this sample came in Nashville, where he was playing behind the awesome Shea Weber-Ryan Suter pairing. It’s nice to see that the Rangers’ second pairing, when healthy, should be effective together when they’re shorthanded. McDonagh comes in slightly negative, which makes me wonder what his numbers on the PK without Dan Girardi look like. For a while now, I’ve thought that Girardi is a giant dead weight on his linemates. This data, plus the 5v5 stats I looked at last week, only solidify my belief that the Rangers made a grave error in giving him that monster contract extension.

Further on Staal, I think the penalty kill is one area of the game where slower, stay-at-home defensemen like him can excel. They aren’t going to drive puck possession, but that isn’t the name of the game on the PK — it’s shot suppression. Guys like Staal may not be the fleetest of foot, but they’re strong enough along the boards to consistently win 50-50 pucks and get a clearance, which is all that’s needed on the kill. If I expand my data to include all defensemen playing at least 300 minutes at 4v5 from 2007-14, the top five players in terms of CA60RelTM are Chris Chelios (in his late 40s), Wade Redden (wait, what?), Greg de Vries (remember that guy?), P.K. Subban (now we’re starting to make sense!), and Steve Montador (…and it’s gone).

Just for fun, let’s take a look at how some former Rangers defensemen do on the penalty kill. I used the same constraints as I did with my analysis of the current Rangers defensive corps: last four seasons, at least 200 minutes.NYR_formerD4v5
sigh
. Damn it, are you serious? I feel like we’re being mocked at this point. Side note: I keep hate-watching Lightning games in order to see Stralman and reminisce/think about what could have been/maybe cry a little. On another note, for all of the grief that Rangers fans (myself included) gave Michael Del Zotto about his defensive zone coverage, he was awesome at preventing shot attempts on the penalty kill. Perhaps the Flyers giving him a contract over the summer isn’t quite as hilarious as we thought. Also, not bad work by Rozsival, especially considering the fact his teammates are superstars.

One final note on this subject: CA60RelTM for defensemen has a lot of volatility. Just doing a quick and dirty regression analysis seemingly showed basically no relationship between one year and the next:

4v5CA60RelTMCorr
An R-squared of 0.049 isn’t too convincing, but we’re dealing with tiny ice times (I had to set the minimum TOI at 100 minutes just to get a decent amount of players), and the overall sample size is pretty small as well (n=88 for this particular two-season comparison). Given those limitations, the R-squared value here makes sense. The next step would be to see how well larger samples correlate. For example, comparing CA60RelTM in years 1 and 2 combined correlates to the same stat in years 3 and 4 combined. That’s a project for another day, though. For now, we’ve got a decent understanding of what a good defenseman looks like on the penalty kill, and how current Rangers blueliners stack up against the rest of the league.

 

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>