View Single Post
01-27-2012, 02:22 PM
Registered User
Join Date: Jul 2004
Posts: 21,123
vCash: 500
Originally Posted by blue425 View Post
Fixed it for you. Hank has been many things in the past six years...but average is not something he has ever been over the course of an entire season.
They mean his average, not to being an average goaltender, but it still doesn't leave any leeway for a guy to improve or have a great year or anything like that, or take into account that maybe this team is better in front of Lundqvist than any team he's had in the past, or that he's playing fewer games which has allowed him to be more consistent and not have as many down periods in his play like he has in the past

Sabremetrics in hockey should not be completely blasted or ignored. They should be analyzed based on what they actually provide. There is a use for statistics in hockey. However, unlike in baseball, where they started creating advanced stats back in the early 70's (perhaps even before that), stats in hockey is in its infancy. Billy Beane built a team with no defense because "defense doesn't matter". This has since been refuted. Blindly stating "we don't need statz in hockey! stick to something else NERDZ1!...stats are stoopid" is an ignorant way to look at what stats provide.

The key here is that stats predict SOLELY based on what has already happened. Players will regress at age 36 because they always have, etc. Overall offense in the league is down compared with last year and even more so than the year before. Perhaps that means that shooting percentages contribute less to wins than they have in previous year. Until shot quality is quantified accurately, goalie skills will never be evaluated correctly. The problem with these stats seems to be that they don't accurately quantify all aspects of the game before making their predictions. If one was to based the Rangers performance based on 1985 offense, one would predict that their GPG average is way too low and that they will suck. Unlike baseball where trends tend to hold over the course of decades, in hockey this is not the case. Power play numbers move back and forth depending on the refereeing that year and which stars are playing and the quality of goalies. Without adjusting for all of these things (similar to Park factors in baseball which account for longer fences or defense metrics which account for balls a fielder "should have" gotten to), I believe the metrics in hockey are still to primitive to be useful.
And this is also a real good post IMO. Things just don't seem fully fleshed out yet and predictions based off of basically unreliable stats analysis seem kind of misleading to me.

I also feel like things like defense and goaltending are way underrated in the current state of advanced hockey stats, possibly because they're not terribly easy to quantify in some ways.

Levitate is offline   Reply With Quote