While the Celtics are an outlier, don't these models usually predict teams to be pretty bunched up without as many high performing teams as you know there will be?I am kind of shocked by the parity predicted by those models. only two teams with > 50 wins. is this some quirk in the models?
My understanding is that they are predicting 50th percentile projections, and that will pull everyone back to the pack. In reality, out of a group of 30 teams, there are inevitably going to be a few teams that hit 80-90th percentile projections and "over perform". But you can't predict which teams those will be prospectively.
Put another way, these models don't actually think only 2 teams till win more than 50 games this year. They just think only 2 teams 50th percentile projection is over 50 wins. Big difference.