I won't start posting the official picks for at least another week - you need at least four weeks of data for the model to really work - but I promised week 4 picks so here they are.
Louisiana Tech. Last year they took Texas A&M to the wire. This year, Vegas likes Kansas by 9 and the model likes Kansas by more than a dozen. The model likes Stanford by less than a field goal; that is to say, the model likes Stanford by less than Joel Stave deciding to lay the ball on the ground so it could get dogpiled and the ref failing to get the ball set in time so the clock expires. It gives Georgia Tech a massive 18 point advantage against UNC, and the model likes UCLA by 50, which is its way of saying UCLA can pick the score.
Have tracked historical performance of the model? I.e., has it been more accurate than the Vegas point spread?
ReplyDeleteThis early in the season I doubt it, but it was for the second half of last season (under CF by the numbers): http://www.thepredictiontracker.com/ncaaresults.php?type=2&orderby=wpct%20desc&year=12
Delete