To be fair xg isn't one stat, it's a variety of different models produced by different companies, all which will give different outcomes.
Your point is generally true though. Although I do wonder, lets say they're using data from top European leagues, would we not expect those players to be better finishers than Championship players, on average? So perhaps these xg figures are overstated for the level.
All the models are proprietary though so it's hard to get real detail on how they're calculated.