Is the Red Wings Penalty Kill Better?

There have been several bright spots in the first five games of the season, and one of them has been the Red Wings excellent penalty killing. The Wings penalty kill has been a perfect 17 for 17 so far, and we've heard a lot about how Tony Granato's PK is a lot more aggressive in an attempt to use the Red Wings speed to their advantage. This is fun and all, but something simple like "17 for 17" isn't super satisfying for me because random fluctuation is a very real thing. If you flip a coin 10,000 times, somewhere in there you're probably going to get heads 7-8 times in a row, even though actually doing so is very unlikely. What I'd like to try to figure out is if the new and improved penalty killing is just coins coming up heads, or if there is a measurable skill improvement happening here.

If you are uncomfortable with the idea of random variation happening, think about what you know of power play goals. You see one roughly 1-2 times a game - how often are those goals purely skill? This power play goal, for example, is purely a skill goal. No weird bounces, no terrible PK work - just raw skill making an amazing play while at 5v4. However, we also know that a lot of PP goals result from firing the puck at the net and praying you get a bounce. The natural question then is how rare is a 17 kill streak? If it's relatively common, there's a decent chance we're just seeing 17 straight PKs where the Red Wings got no bad bounces. If it's very rare, there's got to be some skill seeping in.

Penalty Kill Success Rate

I went through the Red Wings penalty kill results from last season to give us a comparison. The Wings were 12th last year, killing off 83% of all penalties - which is to say they were just slightly better than the league average. As it turns out, during the month of March the Red Wings managed to kill off exactly 17 straight penalties. Now, this was their best of the entire season, but this should cause us to be a little cautious before concluding that killing 17 straight penalties means the Wings are going to be inherently better at killing them off this year. The fact that this year's 17 straight kills comes from the beginning of the season may cause a little more excitement than normal, but the reality is there's no inherent difference between 17 straight in October and 17 straight in March - you can kill 17 straight and still not be a top-10 PK outfit.

The 17 straight kills last year was a season high, but there was another impressive streak in early November where the Wings killed off 28 of 29 penalties - good for a 96.6% clip. In other words, if the Red Wings were to give up a goal on their first penalty of the game against Montreal tonight, they would need to kill off 11 in a row immediately after that to equal their best longer run of last season. At first glance, the evidence seems to suggest it's much too early to conclude that Granato has really improved the Red Wings penalty kill, despite how impressive the first five games have really been.

Penalty Kill Shot Rates

However, there's another way to measure penalty killing success. We can adjust a little bit for lucky bounces by measuring how many shots the PK is allowing. Unlucky bounces are absolutely going to happen while killing penalties, but a team that gives up 25% fewer shots will likely give up fewer goals. This isn't really hard and fast because shot quality is probably more of a factor on penalty killing, but shot rates still matter. So let's look at what kind of shot rates the Red Wings allowed last year versus this year, and see if there's an improvement there. To measure this, I'll use Fenwick, which is all unblocked shot attempts. Including shots that don't get blocked but miss the net seems wise to increase sample size, but using Corsi which includes blocked shots takes away the very real skill of blocking shots while on the PK. So let's go with Fenwick, and we'll adjust it for time on ice - which also eliminates the factor of short PPs (shortened by another penalty, etc).

First off, here is the Red Wings Fenwick against/60 from last year, and I'll include the best, worst, and median team for comparison's sake.

Team Fenwick Against/60
New Jersey (1st) 61.2
Detroit (8th) 67.3
Carolina (15th) 72.1
Washington (30th) 91.7

The basic way of reading this is for every 60 minutes of penalty killing, New Jersey gave up 61.2 unblocked shot attempts. So on average, for every 2 minute penalty kill, they gave up only a tiny bit more than 2 unblocked shots. Detroit wasn't much higher, so the Wings were already pretty good at this last year. There's more to killing penalties than just preventing shots of course, but doing it this way also helps us remove goaltender impact a little bit and focus on the skaters killing within the system. So let's compare that with that same exact chart for the first few games of this year:

Team Fenwick Against/60
Minnesota (1st) 46.9
Detroit (4th) 53.5
Buffalo (15th) 71.1
New York Isles (30th) 110.1

So yes, the Red Wings are doing better so far at preventing shot attempts than they were last year. That's a tangible improvement. However, two massive caveats need apply here:

  1. As you can tell at a glance, the numbers are pretty extreme here with such small sample sizes. Minnesota is hilariously below last year's league leading New Jersey 61.2 number, and a smart person would expect that Minnesota would regress a little bit, even if they really are that good at killing penalties. Detroit is actually one of seven different teams who are preventing shots better than New Jersey did last year, and that should make us very suspicious of this supposed improvement.
  2. The real question is whether that Fenwick against/60 number can fluctuate as wildly as PK percentage.

As it turns out, we can actually test #2 as well. I decided to take that stretch of last season where the Wings killed off 17 straight penalties and see how well Detroit prevented shot attempts. Here are their numbers last March:

Team Fenwick Against/60
Detroit (March 6-14 2014) 31.3

Funny how that works. In terms of limiting shots, the Red Wings penalty kill was actually more successful in 17 kills last March than they have been in the first 17 kills of this year. With this knowledge, it's a bit of a stretch to conclude that the Red Wings penalty kill is inherently better than last season. We can definitely conclude that the PK has been amazingly good in the first 5 games because it very obviously has been, but let's give it some time before we conclude that the redesigned PK structure is actually superior to the one used last year. There's a difference between a hot streak and sustained success, and right now we just have no way of knowing which one we're looking at.