The other day I wrote about what I hope for from hockey analytics in 2016. It largely focused on more investigation into the impact of coaching and specific roles players play have on individual statistics. For the most part hockey analytics has developed by looking at what team statistics correlate well with team success and then transferring those observations to player evaluation (i.e. Corsi correlates with winning so good players should have good Corsi numbers as well). While transferring what we know from the team level to the player level has merit one problem I have is we have largely treated players as a homogeneous group. Sure we often often look at forwards and defensemen separately but we haven’t gone any further than that. We haven’t look at offensive defensemen and defensive defensemen separately or offensive and defensive forwards separately. This doesn’t make a lot of sense to me because in reality, as far as role, usage and statistics are concerned, Erik Karlsson might have more in common with an offensive forward than a defensive defenseman. All of this has led to what I believe is an over-simplification of hockey analytics.
Pretty much every day, if not multiple times a day, I read an article or a tweet that exhibits this over-simplification and it mostly revolves around puck possession. Puck possession in hockey is incredibly important in hockey but I am at the point where I completely despise the term simply because it is getting overused and misused. This tweet is a prime example:
— Sean Tierney (@SeanTierneyTSS) January 8, 2016
Now Sean Tierney is not alone in believing that Muzzin is one of the games best defenders. It is a relatively popular belief in hockey analytics largely because Muzzin has been a great possession (read: Corsi) player. It is good to be a great possession player but it isn’t sufficient to be a great defender and this is the case with Muzzin. I’ll explain but let’s first look at Muzzin’s statistics.
|Season||CA60 RelTM||Sv% RelTM||GA60 RelTM||GF% RelTM|
(The table above are 5v5 close statistics)
There is a lot to be concerned with in the table above. First, Muzzin’s CA60 RelTM has steadily gotten worse to the point where his teammates have worse CA60 stats with Muzzin than apart from him. Making matters worse is that his Sv%RelTM stats have steadily gotten worse. As a result of this his GA60 RelTM stats have dropped off dramatically. Think about exactly what the above table is telling you. This season, on average, Muzzin’s teammates have a goals against rate 1.22 goals per 60 minutes better apart from Muzzin than with Muzzin. Last season it was 0.59 and the season prior it was 0.32. All of this has had a devasting impact on the the goals battle. Kings players are better not playing with Muzzin than with Muzzin. None of this screams ‘one of the games best defenders’ to me even though, until this season, his Corsi had been outstanding.
In many ways Dion Phaneuf is the anti-Muzzin the past several seasons so let’s have a look at his statistics while in a Leafs uniform.
|Season||CA60 RelTM||Sv% RelTM||GA60 RelTM||GF% RelTM|
For the three years from 2012-13 to 2014-15 Muzzin had an average CA60 RelTM of -6.77 compared to Phaneuf’s +7.48. That is a swing of 14.25 in Muzzin’s favour. It is a big difference.
During these three years Muzzin had an average Sv%RelTM of -1.67 (and worse this year) compared to Phaneuf’s +1.56, a swing of +3.24 percentage points in Phaneuf’s favour. To put that in perspective, that is the difference between a .900 save percentage goalie and a .932 save percentage goalie. That’s huge. The net effect is Muzzin had a GA60 RelTM of +0.07 compared to Phaneuf’s -0.26 for a swing of -0.33 in Phaneuf’s favour.
To summarize, while Muzzin has mostly had a negative impact on his teams goals against rate Phaneuf with the Leafs has mostly had a positive impact. There are other factors to consider like quality of teammates but it begs the question, who is the better defender and more to the point, is Muzzin really among the league’s best defenders? Is he even the best defender on the Kings? Not counting the just acquired Luke Schenn, Muzzin has the worst GA60 on the Kings this season after being second worst last season and the worst in 2013-14.
Bringing this back to roles you will notice that Phaneuf’s CA60 RelTM has improved significantly this season. If you are an observer of the Leafs you will also know that Phaneuf’s role on the team has changed. He is no longer the primary shut down defender, his defensive zone starts have dropped, he is getting less ice time on the PK and is playing a more offensive role.
Muzzin on the other hand has seen his CA60 RelTM drop over the past four seasons while at the same time has seen his defensive zone starts and PK ice time rise as a result of being relied upon more in a defensive role.
In summary, as Phaneuf shifts to a more offensive role, his CA60 RelTM improves dramatically. As Muzzin shifts to a more defensive role, his CA60 RelTM drops dramatically. There is a direct link between offensive/defensive role and CA60. Have a look at some other players who have seen their roles change from offensive to defensive or vice versa. Most will show the same trends. All of this shows the importance that roles can play in player evaluation. To properly evaluate players hockey analytics must shift focus from evaluating players as a homogeneous group to one where we evaluate players in the roles they are given. My hope is that in 2016 there is a shift in the hockey analytics community to do just that.