View Single Post
Old
02-10-2013, 05:10 PM
  #86
garret9
AKA#VitoCorrelationi
 
garret9's Avatar
 
Join Date: Mar 2012
Location: Vancouver
Posts: 13,758
vCash: 500
Corsi and Fenwick are two great ways to determine whether or not a player is "beating" it's opposition by looking at whether you out-chance your opponents. Problem I've always have had, is that it disregards WHY you out-chance your opponents. Is it because you're really good defensively (Scuderi), a two-way kind of player (Enstrom), or an offensive monster (Byfuglien)?

The images below are made from David Johnson's stats.hockeyanalysis.com, which is the same site I get the with/without raw data from. He's attempted to answer the question above by separating the events for and against the team when the player is on and off the ice, and putting it through an algorithm to try and diminish team affects... it's neat.

The two problems I have with it is related and both are in the realm of usage, or rather zone starts and match-ups bias. It's a significantly harder task to suppress shots from Sidney Crosby when you line up against him in your defensive zone than it is to suppress shots from Chris Thorburn when you line up against him in your offensive zone. He hasn't attempted to diminish match-up affects, but to diminish zone starts what he does is remove all events that occur within 10 seconds of the puck being dropped. Personally I think it's an improvement but not quite there, as 10s seems pretty arbitrary... meh! Still interesting.

We're currently at a realm of really small sample size but here we go.
The solid line at 1.0 is the point of being a benefit or a hindrance. The Y-axis is his offensive results rating while the X-axis is his defensive results rating, in regard to creating and preventing the respective event (Fenwick or Corsi). In theory, top left is offensive, top right is two-way, bottom right is defensive, and bottom left is a ****ty player.

I wouldn't take much into account quite yet, because the sample size is extraordinarly small and is very much affected by usage (ex: poor Enstrom and Hainsey have been leaned on VERY heavily by Noel while Byfuglien and Bogosian have been out).

Defense



Offense




I'd say give it till game 20, then it should start looking better and more along the lines of what's going on.

garret9 is offline   Reply With Quote