View Single Post
11-21-2012, 03:02 PM
eklunds source
Registered User
Join Date: Jul 2008
Location: Ed Snider's basement
Posts: 7,919
vCash: 500
Originally Posted by MastuhNinks View Post
I don't think you understand, simply looking at GA ON/60 alone is silly and pretty much useless. Which is why the list is dominated by players with good goaltenders. I'm not saying Garrison is a bad defensemen, I'm saying that the article the user used to justify it was very flawed. I actually like Garrison, I think he's one of the most underrated offseason acquisitions, I just don't think he's quite on the level of some of the guys who were omitted from the list. I'll agree that maybe I shouldn't have picked on him like I did with Hedman and Johnson, but my point was never really that those players are bad defensively, just not top 20.

For argument's sake, let's say both of these players in this example have a similar Corsi Rel QoC and Corsi Rel QoT.
Player A has a GA ON/60 of 2.69, and a GA OFF/60 of 3.25.
Player B has a GA ON/60 of 2.50, and a GA OFF/60 of 2.45.

Is Player B really the more impressive defensive player? Should Player A be punished for playing on a team that clearly has some defensive problems, even though his team is significantly better when he's on the ice?

Now, Garrison's GA ON/60 actually is better than his GA OFF/60 by a good amount, but that's not really the point, the point is that heavily team-influenced statistics are useless without context. Just amalgamating a random choice of defensive metrics doesn't make a lot of sense unless there's a method to it (e.x. If comparing Relative Corsi to Corsi Rel QoC, you should also take into account Corsi Rel QoT, likewise GA ON/60 without taking into account GA OFF/60 doesn't mean much).
Goals for/against don't matter.

Well, they do matter, because that's what determines who wins the games, but judging a player individually by goals for/against is an absolutely awful way to approach the situation. They're borderline useless.

Originally Posted by overpass View Post
In fact, [Kuba's] play may have had something to do with the fact that Brian Elliott had the worst SV% in Ottawa and the best SV% in St Louis.
This is why we don't make up stories or narratives to fit the data - they're usually wrong.

Kuba in '10-'11 (even strength)
Time periodKuba's on-ice sv%Elliott EVSV%Leclaire EVSV%Anderson EVSV%Lehner EVSV%Brodeur EVSV%
First quarter of season0.8840.9140.859   
Second quarter of season0.8770.8920.957  0.778
Third quarter of season0.8730.856 0.9520.8800.818
Fourth quarter of season0.911  0.928  
(McElhinney not included)

If Kuba's on-ice shot numbers were the same, it may owe as much to the fact that he played in a similar role with similar players as it does to his individual play.
Top 5 defense partners in '09-'10:
1) Karlsson (327min)
2) Phillips (207min)
3) Campoli (139min)
4) Lee (33min)
5) Volchenkov (17min)

Top 5 defense partners in '10-'11:
1) Karlsson (428min) +101min
2) Butler (147min) +147min
3) Lee (131min) +98min
4) Gonchar (127min) +127min
5) Campoli (66min) -73min

He played less than half as much with Campoli, had never before played with Gonchar or Butler, quadrupled the time he spent with Lee, added over a hundred minutes with Karlsson, and played 160 minutes less with Phillips.

I understand that plus-minus may not be very meaningful for evaluating Kuba's performance but I'm not sure the shot plus-minus metrics are either. He's a supporting minutes eater who is only one of five skaters on the ice.
I'm not saying his shot +/- is 100% driven by Kuba. I'm saying that his performance from year to year was very similar, and the %s that are out of his control (how well the opposing goalie and his own goalie played when he was on the ice, versus how well they played when he was on the bench) were wild. I picked Kuba as an example because they're pretty radical differences.

Also, I'm not sure what you mean by "supporting minutes eater"... He played 16:06 per game at even strength, plus 2:29/game shorthanded and 2:07/game on the powerplay. "Supporting minutes" players don't typically average almost 21 minutes per game.

eklunds source is offline   Reply With Quote