Updated Player Rankings

Over the past week I have found some free time to update the player ranking algorithm. Not only have I improved the algorithm but I have also fixed a bug/flaw in the code which overly penalized players for giving up short handed goals. The end result is that players who play a lot on the penalty kill should see their defense ratings rise, sometimes quite dramatically. I know there was concern about Lidstrom’s (and others) poor defensive ratings last time so now they are fixed. Lidstrom now gets a very respectably 1.42 defensive rating. You can view all qualifying players rankings by clicking on the appropriate links in the menu on the left.

You will still find some ‘strange’ results in the rankings, like Kobasew’s 3.96 defensive rating, but these will fall off as the players play more ice time. Kobasew gets the high defensive rating because he has had a league low (of players with 350 minutes of ice time) 3 goals scored against his team while he has been on the ice. That’s pretty impressive but probably has more to do with luck than his raw defensive ability. Most of the odd rankings you will find in my rankings can also be found in the rankings recently created by JavaGeek and JavaGeek uses a fairly different algorithm than I do so it appears that those oddities aren’t due to flaws in the algorithm. Generally, ratings significantly over 1.50 are probably not sustainable but overall I am fairly pleased with the results.

This article has 7 Comments

  1. Ok I was just looking over your rankings for the Leafs in specific, and I don’t quite get a few things. Firstly, how is Hal Gill the highest rated offensive contributor, and lowest rated defender of their top 4 D men??

    The guy isn’t an “offensive” defenceman. Perhaps his contribution to the offense by providing a stable back end is substantial but considering you’ve got him rated as a sub-par defender I find that a tad strange. Frankly I just can’t wrap my head around Bryan McCabe having a lower offensive rating and a higher defensive rating than Gill. It doesn’t really make much sense.

    Similarly, how does Zdeno Chara get a 1.01 defensive rating while Paul Mara gets a 1.03?

    I’d really like to know how you’re calculating the “expected” production values for the players on the ice. For instance it doesn’t make sense to me that Teemu Selanne, Chris Kunitz, and Andy McDonald would all have such high offensive ratings since they are all on the same line. If they’re ALL getting such a high rating that would imply they’re all better offensive players which should increase expected production, which would thus reduce the impact of any one of them in particular. So unless they’re magically all having the same impact at the same time for some weird synergy reaction I don’t know how you can break out their impact individually.

    I think when all is said and done, I just can’t accept that you have an algorithm that is isolating individual player play enough to make these numbers meaningful. They’re all interdependent, and while you might be pulling some theoretical values out of all this, I don’t know that it’s attributable solely to their “offensive” or “defensive” ability. I think you’re likely reifying things a tad too much into amorphous concepts of offensive and defensive contributions.

  2. Like I said, there are some things that don’t quite make sense on the surface, but there are always numbers to explain it.

    Hal Gill: Gill is one of those lucky guys who has been on the ice when a lot of goals have been scored. I would suspect his offensive rating to drop as the sample sizes (his ice time) get larger. But here is an interesting stat. Gill has 8 even strength points. McCabe has 9.

    Chara: Chara is -3 on the season and is a major component of the 4th worst PK team in the league. Last year he had the worst +/- of all of Ottawa’s regular defensemen. Could it be that his defensive ability is over rated?

    Selanne/Kunitz/McDonald – It is reality that if 3 players always play together it is difficult to isolate their individual contributions. It is a limitation of the (and any) algorithm. The more they play apart the better the algorithm will work so as the season goes on and they get more and more minutes apart things should separate out a little.

  3. Chara also led Ottawa in Minutes played per game by almost 4 minutes. He also led the team in SH Time on the ice. That’d be the team with the 3rd best penalty kill last season, and the 2nd best goals against total in the league. That stat would be a little strange if he was an “over-rated” defender.

    Also you might want to note that while Boston has the 4th worst PK in the league right now, they’ve taken the 10th fewest penalties in the league so percentage wise, giving up the 6th most power play goals against might be a bit of a problem. He’s on the ice for a minute more than the next closest penalty killer for the Bruins, and 7 minutes more than the next closest player. He’s obviously not the SOLE reason they’re giving up so many goals. In fact you might want to recall the fact that Boston’s Goaltending situation was pretty scary over the first few months of the season. In October Tim Thomas had a .899 Save Percentage, and now in December he’s sporting an awesome .889 Save percentage. In October Toivonen had a wicked awesome .884 save percentage. Stellar goaltending sure makes D look great.

    As for Hal Gill’s 8 even strength points and McCabe’s 9… that doesn’t exactly explain Hal having a 1.48 to Bryan’s 1.14 rating. Especially considering Gill plays only 1 minute less at even strength than McCabe does a game. The fact that McCabe plays about 6 and a half more minutes on the PP than Gill might explain the distinction in their offensive value to the team. Your numbers indicate that Gill would be better used on the PP than McCabe, since he’s a more noteworthy offensive contributor. Plus… how many even strength goals against has McCabe been out there for in comparison to Gill? Because I don’t think the 1.09 to 0.88 defensive rating makes much sense either.

  4. Don’t misconstrue my discussion around McCabe and Gill as me being a large McCabe supporter, frankly I dislike him intensely and think he’s one of the worst defenders on the team. That’s actually why I find it odd he’s ranked more highly from the defensive perspective.

  5. Like I said, there are a a few odd results but that doesn’t mean the algorithm doesn’t work or should be ignored. Gill’s offensive rating is one of those odd results.

    As for Chara, I am still not convinced that he is a star defensive defenseman. I think he can be beat by speed. He had a horrible playoff series against Buffalo last year. And last year being the lowest +/- defenseman on a team where getting +’s was easy all while playing by far the most even strength minutes should tell you something as well.

  6. I don’t understand how playing more minutes automatically translates to more +’s.

    Plus you could also make note of the fact that at home he was tied for 5th on the team and had the 2nd best total of all Ottawa D at +13.

    Meszaros was +12 and Phillips was +10 at home. On the road they happened to be +22 and +9 respectively.

    Based on last season’s numbers you should be wholeheartedly convinced that the REAL reason Ottawa was so good on D were Wade Redden and Andrej Meszaros. This season, Meszaros is a wonderful -14. Redden has played a grand total of 20 games, has 7 points and is a -1.

    Chara on the other hand has played 33 games, has 22 points, and is a -2, which is pretty good considering he was -7 after the first month of the season, which is when Boston’s goaltending was most atrocious.

  7. Oh and in regards to his series against Buffalo in which he was oh so atrocious, he was -2 for the series. Redden was -7. Meszaros was -3. If we’re thinking he was “exposed” by speed wouldn’t it make more sense for him to be the one with the horrible +/- in that series? Oh and before you jump on him for his PK, Buffalo only scored 3 powerplay goals in the entire series.

    He was on the ice for 2 goals for and 2 against in the first game, 1 power play goal for each team. In game 2 he was on the ice for all 3 goals in a 2-1 Buffalo win. All were at even strength so that made him -1. In game 3 he was on the ice for 2 Ottawa power play goals, and no goals against. He wasn’t on the ice for any goals in game 4. In game 5 he was on the ice for 2 goals. 1 Power Play goal for the Sens, and 1 Power Play goal for the Sabres.

    Overall that doesn’t sound like he was “horribly exposed” by speed. Especially when you consider the comparison to Redden’s stats. If you want to look at the game sheets it might also serve as a reminder that it wasn’t like Ottawa gave up a boatload of goals in any of the games. They lost 3 of them in OT, and the only one they lost in regulation was 2-1. That’s more crap luck than anything… and believe me I hate the Sens, I’m a Leafs Fan. But be realistic. Chara is an awesome defender and to insinuate otherwise is a bit silly.

Comments are closed.