View Single Post
Old
05-17-2013, 01:03 PM
  #640
I am the Liquor
Registered User
 
I am the Liquor's Avatar
 
Join Date: Oct 2008
Location: Sunnyvale
Country: Canada
Posts: 34,284
vCash: 7000
Quote:
Originally Posted by misfit View Post
I don't think the parameters changed, it's just that the way they calculate it requires a ton of events for the numbers to start making sense. Because of that, the QOC numbers are usually worthless until enough data is collected. But over time, things even out and it's actually fairly consistent with what most people who pay attention to who players are on the ice against would tell you. You'd probably get two very similar lists if you were to ask a season ticket holder who didn't own a computer, and a guy who didn't know a thing about hockey but knew where to find QOC data online, to make a list of every Oiler player in order of the quality of opposition they faced.
It absolutely did change. It was originally based on +/- I believe. Hence Cooke= tougher competition than Malkin in the one year I looked at (a few years ago). It was then changed to relcorsi or something like that. Not sure if that corrected the problem or not. The lack of credibility made me lose interest tbh.

Quote:
Viagra was originaly invented to treat high blood pressure, and Play-Doh was developed as a cleaning agent.

Sometimes things have uses beyond what they were intended for. And in the case of hockey and Corsi, it seems pretty reasonable that a goalie's workload is at least in part tied to the quality of the players in front of him (in fact, I'd say they're very much connected).
The problem remains that they are taking a team generated stat and trying to apply it to individuals. The white noise pretty much muddies anything of meaning. You can use those stats to quantify how team A is doing when compared to team B, but as soon as you try to assign individual values you are lost in a sea of uncontrollable variables and the exercise becomes mired in futility.

I do think that GF/60 and GA/60 are more meaningful/useful than straight +/-, so its not like it is all garbage, but a lot of the other stuff is not useful imo.

Just look back over the last four years or so. The hot new stat of the day comes along, then is shortly forgotten about when a new hot stat comes along. There is no continuity. Then there is the constant rejigging process that starts out with Corsi, which becomes relcorsi which then becomes adj/relcorsi.

Its actually quite amusing.

To me it seems amateur. No scientist or mathematician would ever sign off on what is going on there.

I am the Liquor is offline