Quote:
Originally Posted by Fighting Banana Slug
Is it such that the model average error rate is 10.4 in either direction? If so, that seems pretty useless. My "model" would take the middle 20 teams and give the exact same point total from the prior year. Top 5 teams I would deduct 10. Bottom 5 teams I would add 10 for regression to the mean and take my chances with that.
|
It's really 5 wins in either direction of variance which I think is probably fine. I know rookies, goaltending and injuries can really swing the standings over the course of a season.