The Leafs' Penalty Kill Isn't Broken - Your Expectations Might Be

Editor's Note: All numbers are as of Saturday morning.

We're 22 games into the season - a season that saw many Leaf fans enter with high expectations for this year and many fans of Advanced Statistics with low ones. The reasons for those low expectations on the part of those of us that look into the analytics and underlying numbers behind the game are many and I won't delve into all of them in the posting.

Many of the most regularly repeated refrains surround the concepts of sustainability or reliability. A large proportion of Leaf followers have spent much of the past year or so touting the vast improvement on the PK as a sign of the coaching genius of Randy Carlyle - ignoring the fact that it's an amazingly unreliable ranking year over year.

In addition, not only is the ranking of PK% unreliable, in the grand scheme of the NHL standings, a team's penalty kill is far from a key ingredient. The top team in the NHL right now - Chicago - sits 29th in the PK% rankings... and yet they're first overall in team Fenwick For % at 5v5 Close (which is far more integral to a team's overall performance).

Hanging your team's hat on their PK is unlikely to be worthwhile in the long run. They're unlikely to stay there, and even if they do - it doesn't mean a whole heck of a lot. Jonathan Willis took a look at whether the re-alignment of the Leafs' fortunes had already begun and the penalty kill was one area which he highlighted.

So let's look at a few things that are relevant to the Leafs PK in recent years and so far this season. First - an exploration of two of the key features of a good PK. Goaltending and shot attempt suppression. These two components should make obvious sense as SV% is the percentage of shots stopped, and shot attempts are the events that end up in the net.

Goaltending

Many of us have heard the old truism "your best penalty killer is often your goalie". This is not without cause - as in most of the top PK units, the goaltending is outstanding. In fact - over the last 6 seasons - 4 of the top 5 teams in GA/60 while down one player are also ranked in the top 5 in SV% while down a man. Conversely, over that same stretch of 6 seasons, 4 of the 5 worst teams in GA/60 while down one player ranked in the bottom 5 in SV% while down a man - with dead last being your very own Toronto Maple Leafs. Obviously goaltending is key to the PK.

Unfortunately season to season 4v5 SV% isn't that reliable. Year over year the correlation is only 0.0612. While that is significant, it implies that it is far from a certainty that a team will repeat it's numbers from the previous season. This result implies that we should regress a team's 4v5 SV% by around 94% towards the mean (0.877) to estimate results for a subsequent season.

So if we were to do that - based on the Leafs 4v5 SV% of .897 last year - we should estimate that they would post a team SV% of .878 this season. Last season James Reimer posted a .914 SV% at 4v5, but he is sitting with an .878 so far this season. Currently the Leaf's team 4v5 SV% is sitting at .907 - thanks to Bernier's .921 - which is still higher than last year and leaves lots of room for regression. My advice would be to prepare for it before it gets here.

Shot Suppression

This may be where the most alarming change has arisen for the Leafs this season. Were it not for the abnormal improvement in the team's PK SV% (which again - is unlikely to continue) the change in shots allowed against on the PK alone would be serious cause for concern. Though less integral to a top PK than goaltending - Shots and Shot Attempts faced on the PK do correlate with team success.

From 2007 through 2012 the Leafs ranked 30th overall in the NHL in PK Success allowing 7.98 GA/60 while in 4v5 manpower situations. Their team SV% during this stretch was an atrocious (and league worst) .842. In addition to this, they also managed to rank 25th in the NHL in Fenwick Events Against per 60 mins of short handed ice time. Unsurprisingly - when you combine the worst goaltending in the NHL with the 25th best team defensive structure in front of those goalies - you end up DFL in penalty killing.

Last year they suddenly improved... why? Well for starters they jumped to 4th in the NHL in Team SV% at 4v5 with a .897. In addition to this they rose to 9th best in Fenwick Events Against per 60 mins (61.8 FA60).

Ironically this isn't the first time in recent memory they ended the year ranked 9th in this category. The last time was 2010-11, under Ron Wilson (oft ridiculed for his views on fronting in the DZ). The key distinction in 2010-11 was the astronomically atrocious goaltending that posted a .831 SV% while down a man.

So how's that shot suppression thing going this year? Not so great. What does "not so great" mean? 10th? 11th? How bad is it? 23rd... the Leafs have gone from 9th to 23rd in FA60. Although - again - is this abnormal as variations go? Well let's look at the team's ranking over the past 7 seasons to explore this:

21st, 24th, 28th, 9th, 22nd, 9th, 23rd...

Well it looks like they've consistently been in the low 20's with a couple of blips up to 9th. I next considered whether this was a personnel issue and looked at who the top Leafs to play at 4v5 were season to season... and what I noticed was interesting on a few levels.

Look at who the top 4 TOI guys have been over the last 7 years and think about who was brought in as a defensive fix or upgrade for the team:

Why does it seem impossible for the coaching staff to regularly select decent defensive players? Aside from Gunnarsson and Phaneuf who both appear in those numbers over 4 separate seasons each. For instance - here are Nikolai Kulemin's FA/20 numbers at 4v5 over the last 4 years in reverse order starting with this season:

20.329, 19.315, 22.708, 20.208.

Those numbers are very impressive. How impressive you might wonder? Impressive enough to rank 18th amongst all NHL forwards with 200+ 4v5 minutes across the 3 seasons leading into this one. The next best Leaf forward? Jay McClement at 58th... and yet Leaf fans are continually berated with the idea that Jay McClement is the Leafs top defensive forward. Kulemin's peers in this regard are Maxime Talbot, Zach Parise, Joel Ward, Craig Adams, Patrik Berglund, Vladimir Sobotka, and Dustin Brown. These are top penalty killers on top NHL teams over the past 3 years (i.e. Philadelphia, Minnesota, New Jersey, St Louis, Pittsburgh, and Nashville). Additionally, Kulemin is posting numbers like that playing for the LEAFS.

McClement's 58th place ranking might seem impressive until you recognize that his 22.556 puts him at the 60th percentile. He's better than 60% of NHL forward penalty killers. Ranking 18th is the 88th percentile... that means Kulemin is better than 88% of NHL forward penalty killers. Which guy would you rather have playing the most minutes on your PK?

Oh and while Kulemin's been injured early this season - his FA/20 at 4v5 this year? 20.329. Yeah - that ranks 10th in the NHL amongst all forwards with 20+ minutes on the PK so far this year. McClement's FA/20? 28.831 ranks 87th. There are only 105 forwards that have played 20+ mins at 4v5 this season. McClement's currently sitting in the 17th percentile - the vast majority of forwards are posting better numbers than he is. That isn't to suggest McClement is the primary cause of what's ailing the Leafs PK - but he likely isn't the solution either... no matter what the pundits would have you believe.