Being Bad on the PK Means You're Bad Defensively... or Does It?

Hello hockey fans and welcome to my new blog, Bleeding Blue and White.  While this is my first blog post, on my first self-run blog, I have been posting for a while over at PPP under the same name (BCapp).  Thus if you enjoy this post please check out my page over there, for some of my older work.  Now for my first post...

I recently had a discussion with other hockey fans where someone asserted that the Leafs under Wilson have been horrible defensively.  I decided to look into the Leafs actual production under Wilson and I found that they have actually consistently improved their GA at 5 x 5, but their PK has continued to be horrible.  I decided to analyze what the relationship is between total GA, GA at even strength (ES)*, and PK%.  If this sounds interesting to you, continue after the jump!

*Note I used 5 x 5 statistics to represent ES.  Technically ES can include 4 x 4 and 3 x 3 which I did not include.

To analyze the relationships between these three measures I collected them for every team from the 2008/2009 season to today (data over here).

First lets see how well PK% correlates with Total GA/GP:

Apparently it correlates a little bit.  With an R-squared of 0.367 it has a moderate correlation, but it definitely could be stronger.  Lets have a look at how even strength GA/GP correlates with total GA/GP:

This is a pretty high R-squared value of 0.734.  This supports the pretty obvious conclusion that a team's over all defensive game is highly related to their defensive game at even strength.  Finally lets have a look at the relationship between PK% and ES GA/GP:

Apparently it doesn't correlate very well at all.  With an R-squared of 0.092, we have an extremely poor relationship (nearly chance).  In other words a team's defensive game at ES is barely (or potentially not at all) related to their ability to kill a penalty.  This may seem counter intuitive, but it suggests that killing penalties is a specific skill.

So what do I take from all of this?  It appears to me that judging a teams defensive game by their PK is a bit disingenuous.  While a team's skill on the penalty kill is related to their overall defensive game, it is by no means a great measure and is, in fact, a worse measure than their play at ES..  Further, I think it suggests that being good on the PK involves different skills than just being strong defensively as teams that are good defensively at evens can have bad PK's and vica versa.

So the next time someone knocks a team for being shitty defensively because of their bad PK, think twice.

Cheers