How accurate is xG?

And I suppose it may be more accurate in respect of a group of 20 teams, than in respect of any one team? It would be interesting to know what our xG would be if you used the data of the bottom 10 teams - presumably it would go down, because they're less likely to convert any given chance? So actual and expected would converge.
Or something.

Posted By: paulg, Oct 7, 13:07:02