Not fully having a crack at you, but if there are big differences between a predictive model like xPTS and real league position for a majority of clubs ... the conclusion isn't usually that it's reality that's wrong.
The idea we're overperforming on results feels right, the proposition we should be twelfth doesn't.