Expected goals is not a real thing only made up theory.
I would expect whoever scores the most to most likely score.
Anybody adding some other sort of stat into it is looking to appear smarter than the rest.
What you’re saying “should” be true over a very large sample size, and the fact that it isn’t *exactly* suggests there is room for the model to be tuned up. However, it’s still a useful lens to examine.
If a guy has a career high or a career low Gf in a season, or a under a new coach, or with new line mates, but is maintaining his usual xGF, you can probably infer that the situation or luck is to blame/credit, and that he can regress to the mean if addressed/waited out. There are probably a ton of practical applications of this. But “generating scoring chances” is worthwhile to track, even if it doesn’t always perfectly correlate to generating scoring.
IMO guys who regularly exceed or match their xGF are exactly the type we should be targeting to play with Hyman and McD. You know, scorers. Finishers. Shooters. Real talent.