GDT: Game 24 - CBJ vs PHI - 27 Nov, 7pm, @Nat

Fair enough. I'm no expert, this is just the bits I know.

There are several competing xG models. What they have in common is that they all outperform goals and corsi as predictors of future goals.




Shot type, including backhand/forehand is tracked and included in all of the models. Shooter talent is in some of the xG models and not in others. Everyone agrees it's important on the individual level but it doesn't necessarily improve the predictive power of the model. I don't know which models include it, there's good discussions about incorporating shooter talent here. This discussion on incorporating pre-shot movement (where the pass comes from, rebounds, etc..) might also be interesting.

 

Corsi seems to be a very poor predictor of results due to the fact that teams when behind are going to take more shots than leading teams. Corsi adjusted for game score has to result in better predictive results.

I appreciate the info. I'll look at this later when I have some time to give it.