couple things about xg from corsica's writeup on it
The R^2 value obtained from the relationship [of goals scored to xG] (0.750) significantly exceeds that of goals scored versus unblocked shots (0.586) and even shots on goal (0.619) despite the disadvantage of ignoring whether shots missed the target.
This is referring to the xG of individual players. All of these metrics seem to do much better for individual players, but they fail on a team level when forced to account more for tactics and player interaction.
View attachment 170291
its not perfect but seems to show that xgf correlates more with gf than corsi/fenwick/sog
The fact that the difference between the shot quality metric and the volume ones is around 0.04-0.05 is pretty damning of the shot quality metric - that's definitely not a significant improvement. It honestly pains me to tear this stuff down, because making a shot quality metric like that used to be one of my dreams. I'm in a masters program for statistics in large part because hockey analytics got me interested in the field. The more I learned about stats though, the more I realized how little predictive value hockey analytics had. Kind of a "never meet your heroes" sort of thing.
Edit: I should've read a little further, because Perry himself acknowledges that his model is not predictive.
"This idea of projecting future outcomes is of great importance in analyses relating to hockey and indeed a great number of fields. In its present condition, 5v5 xGF% is not a better predictor of future 5v5 GF% than CF% at the player level. Regular skaters’ 5v5 xGF% in one >400 TOI season did not yield a higher correlation with the next season’s 5v5 GF% than the same test performed with CF%.12 The same variance observed in early shot quality analyses prevents on-ice xG from predicting real goals, or itself for that matter, in any practical way. Though
descriptive of shot quality, the xG model has not yet shown to be appreciably
predictive of future shot quality or goals at the on-ice level.13"