Nucker101
Foundational Poster
- Apr 2, 2013
- 21,189
- 16,686
Yeah, I get that(trust me, I've looked into WAR, GAR, xWar, xGar). It still seems like more of a stat that's trying to condense those advanced stats player cards into one nice number for simplicity sake. I don't think it's completely useless, but I do take it with a grain of salt.Like @ahmon mentioned above, GAR is partly a rating of a player's talent level and partly a description of what happened (whether the player is the reason for it or not). In the case of something like Kase (or Jared McCann this season), a lot of goals went in for the player's team when he was on the ice. Now, given that we're talking Kase and Jared McCann, I don't think we really believe they're the best offensive players in the league. But they had inflated, unsustainable on-ice sh%s and the goals went in. The model sees goals went in and credits them for it (they were there when the goals happened!). It's not really a flaw of the model..it just is what it is. Partly a measure of what happened. It's more complex given it's long regression controlling for many other variables but...I think this is the gist.
Now, Evolving Hockey does have xGAR, which weighs shot metrics rather than on-ice goals. Jared McCann does not rank as highly by this (although still in the top-50.) More reasonable there.
Seth Jones does not rate highly in either GAR or xGAR. I've seen some video work on his slow turning/pivoting. It leads to leaving bad gaps and not "attaching" to his check effectively in the D-zone. It might be related to his multiple ankle injuries, so his decline might not be related to the "Torts effect" and is more about a real lack of mobility. Take that for what you want, but I think there's a huge buyer beware sign on him.
I actually think xGAR works better for team evaluations, but even then I prefer MoneyPuck's Flurry/Score adjusted XG% when looking at team performance, I think that formula has adjusted for some flaws in systems like xGAR.
Personally, I prefer looking at the stats all separately. Zone exits, zone entries, zone denials, shot quality for/against, shooting talent, quality of line mates, zone usage, shot chart heat maps, on-ice sh% for/against, and so on.