NYR Top Penalty Killers - PART I - Forwards

The Rangers #fancystats at 5v5 have, aside from Kreider/Stepan/Zuccarello, been terrible. But one bright spot so far has been the special teams, especially the Penalty Killing.  What was a major mess in 2015/16 has become a team strength in 2016/17.  There are probably many reasons (the bottom 6 forwards signed in the off-season, improved play from the returning skaters, Dan Girardi being a magical broken clock that's on time more than expected) but who has the time to do a deep dive into that???

One thing my Eye Test suggests about the Penalty Kill is it seems to generate more offense than previous seasons.  Sure, Rick Nash was always good for some shorthanded highlights, but this year, the trio of Grabner, Hayes and Miller seem to be getting shorthand breaks and odd-man rushes every game.  [Nash too, though as you'll see, he's missed too much time to make the cut].  If that was true, I'd expect to see players who have strong Corsi and Expected Goals (strong being relative, since this is the PK after all).  As of 1.16.17, Hayes is tied for the league lead with 4 shorthanded points, and Miller is tied for 6th with 3.  [Fun fact: Dan Girardi is tied for 36th with 1].

I pulled from Corsica.hockey a list of all Forwards with over 50 minutes of 4v5 time in 2016/17- a total of 136 skaters.  Nothing scientific about that - 50 minutes just seemed right as a little over 1 PK minute per game to this point.  And 136 skaters has a nice fit for a league where teams probably have a main "Top 4" rotation of penalty-killing forwards.  It's not perfect, as the NYR actually have 5 forwards who clear that 50 minute 4v5 mark:  Grabner, Fast, Hayes, Miller, and Stepan [in descending TOI order], and if Nash was healthy, he would have made it too.  So the NYR really have a Top 6 set of PK forwards - it would be interesting to see if that is somewhat unique, and a possible factor in their success.

I then took a selection of #fancystats from Corsica that suggested both "defensive" and "offensive" results.  Again, nothing scientific here, but for "defense" we have:

  • Corsi Against (CA),
  • Corsi Against per 60 (CA60),
  • Corsi For% (CF%),
  • Relative (to team) Corsi For % (Rel.CF%),
  • Expected Goals Against (xGA),
  • Expected Goals Against per 60 (xGA60),
  • Expected Goals For % (xGF%),
  • Relative (to team)  Expected Goals For % (Rel.xGF%), and

Obviously the percentage metrics are results for both ends of the ice, but since the PK is primarily a "defensive" game state, I put them here.

For offense, I used:

  • Corsi For (CF),
  • Corsi For per 60 (CF60),
  • Individual Corsi For (iCF), 
  • Individual Corsi For per 60 (iCF60), and
  • The same categories for Expected Goals.  

Blahblahblah.  Here are some current results (through 1.16.17) for these Top 5 NYR PK forwards, compared to their peers in that 136-member group.  All data is score and venue adjusted.


I think this is easy to understand, but for example, Derek Stepan has been on the ice for 75 Corsis Against, ranking him 18th of the 136 forwards in the group (the 87th percentile).  As you can see Stepan's been very successful on-ice keeping shots down, and pushing play to the other end (21.88% CF% is 3rd best, 98th percentile).  The rest are mixed results - mostly positive, with some work to be done by Hayes/Miller in getting the shots against down.

Much more fluctuation in Expected Goals Against. Miller has the worst xGA of the entire peer group. Interesting that Stepan is pretty bad too - suggesting while shots against while he's on the ice are low, more of them are "high quality".  Fast, on the other hand, has middling Corsi results, but outstanding Expected Goals Against results.  

Obviously these aren't all the same, but for fun, when you average all these "Defensive" metrics together, you see a lot of "B" level results.  


Hot diggity, that's a lot of blue!  Blue is Good!  We love Blue!  YOU'RE MY BOY BLUE!  So when this group has a chance to shoot the puck on the PK, they do.  NYR have 3 of the top 10 in Individual Corsi For per 60 in this group.  

Expected Goals For is more jumbled, and that's somewhat surprising to me. It *seems* like these guys (especially Grabner and Miller) are getting quality chances regularly, but it may not be the case. Lots of Corsis from the previous chart, but they aren't generating a ton of Expected Goals of those shots.  There's probably an element of sample size here - again we're only talking about 50+ minutes of PK time - a few "quality" shots here and there across the league can change these lists dramatically.  Fast is 9th percentile in Individual Corsi For per 60, but 71st percentile in Individual Expected Goals For per 60.  

As for Hayes Individual Expected Goals?

Well he kinda does look like the Frat Guy who'd never go to class because he's stoned 24/7. 

Taking the average of all these ranks, again a fair amount of Blue, mainly for their Corsi For results.

Looking at Corsica's heat maps for shots and goals (on the "PK", as opposed to 4v5).

I think these look like you'd expect from the number results - Grabner and Fast are a main pair, so they have similar results.  Miller's map looks like a condensed version of Hayes, which makes sense as they are also a pair, but the variation may also suggest an issue with the recorded locations.

There you have it.  These guys have pretty good results on the PK so far.  We probably need to look at the Defense next. I'm gonna take a wild guess that when we do, it'll be like we joined the early 2000s rock band Chevelle...

...Again I'm right in my analysis.