Source: Roy Luck  http://www.flickr.com/photos/royluck/5203755227/in/photostream/

Source: Roy Luck
http://www.flickr.com/photos/royluck/5203755227

 

I’ve updated the Total Hockey Ratings (THoR) for the 2013-14 season.  Both the Even Strength and the All Events can be found on the THoR Page.  Recall that THoR evaluates each play from the NHL’s RTSS system for it’s likelihood to lead to a goal in the subsequent 20 seconds.  We then get a rating for players by accounting for their impact after adjusting for quality of teammates, quality of competition, zone starts, score effects, etc.  The outcome metric here is Wins above replacement (WAR) relative to position.  Details on the methodology can be found on the THoR Page.  Below are some of the highlights from these results from the posted files which include the top 50 players.

http://marionjensen.com/2005/05 Even Strength

Some interesting results here.  I’ll start with EV first.  The top ten here are A. Kopitar, M. Niskanen, M-E Vlasic, R. Suter, A. Sekera, J. Pavelski, J. Jagr, L. Couture, J. Hudler, T. Vanek..  Sekera and Hudler might be a bit of a surprise but none of the others would seem so.  Suter has not always been so highly ranked by THoR but this year he is evaluated well. Couple of other notes on this.  At even strength, THoR is a big fan of Gustav Nyquist of the Red Wings.  And THoR also likes Tampa Bay’s Ondrej Palat as the best candidate for the Calder Trophy but he is.closely followed by Hampus Lindholm of the Ducks. The total impact of Palat and HIndholm is probably very similar but Palat was on the ice for fewer events.  MacKinnon is not rated as highly.   (Note that THoR does not take account of PDO as part of it’s evaluation of players so Nyquist’s high shooting percentage is not relevant to THoR but his high rate of shots per time on the ice is.)  

In this analysis, we also find that the bottom five include: A. Ovechkin, A. Edler, A. MacDonald, N. Grossman, N. Yakupov.  Again, these are not the worst rated players in the league but those that have cost their teams the most due to the amount of ice time they are getting in the THoR methodology.    

Nepomuceno All Events

The top ten evaluating players based upon all events is:  R. Suter, C. Kunitz, J. Jagr, M.Niskanen, M. Giordano, A. Markov, T. Brouwer, J. Carlson, D Doughty and O Ekman-Larsson.

Seven defensemen are on that list and that is due to the fact that we are doing two things.  One, we are  normalizing average player value to the position and, two, we are taking that average relative value and multiplying by the number of plays for which the player was on the ice.  Since top D tend to play more minutes, they are giving more value under this assessment.  This, in particular, helps Suter who was on the ice for just about 25% more plays than Giordano.  On a per play basis Giordano had a greater average impact but Suter had a bigger overall impact on the season.  As with any statistical metric, we can derive variability in these estimates. 

Hampus Lindholm, a defensemen for the Anaheim Ducks,  is the highest rated, 11th, rookie on the list.  He is not a Calder Trophy finalist.  The finalists are Ondrej Palat, Tyler Johnson and Nathan MacKinnon, all forwards, who are rated 38th, 75th and 94th, respectively.  

Again, we see some of the usual names at the top of the THoR list.  Doughty, Steen, Hossa, Hornqvist, Couture, Karlsson, Kopitar and Subban.  THoR is a two-way player metric and so this is not a surprise given the high year to year correlation for THoR.

Of note is that Ovechkin (not in the Top 250) compensates somewhat for his really poor even strength play with PP play that makes him a replacement level player for this year though 

At the bottom of the ratings (not in the file for download), are  A. Edler, R. Regehr, A. Ference, N. Nystrom and J. Cowen.  Again, they were not the worst players but by virtue of their amount of ice time, they hurt their teams the most.  

Also of note here is that Kunitz gets very high marks and higher marks than Crosby.    This is a results of multicollinearity in these data with Crosby and Kunitz playing nearly 4/5th of their 5v5 ice time together (Source: www.behindthenet.ca).  While THoR uses ridge regression to deal with this, it is currently optimized to provide high predictive reliability.  

Alfredson, Jagr,  Markov, Moulson, Niskanen,  Timmonen, Vanek and Vrbata are all in the THoR Top 100 for 2013-14 and they are all UFA’s this summer.  They should have plenty of good offers this offseason (except for Jagr who has resigned with NJD).  That, along with a higher salary cap, should make for an interesting summer. 

One interesting note for long-time hockey analytics followers is the appearance of Sean Couturier and Mark Scheifele in the top 250 players, though Couturier is worth over one win more than Scheifele this year based upon THoR.  Also, David Perron of the Oilers, who was obtained in a trade from the Blues, shows up in the top 100.  

Note that we use a different THoR model for Even Strength than for All Events that accounts for differences between 5v5, 5v4 and 5v3, among others, .