My NFL totals model is showing an extreme bias toward overs this year and has made me pull back from using it at this point. From 2002 to 2010, my model showed a small bias of 0.55 toward the over, but this season that bias is over 1.50. Therefore, the model is picking mostly overs and virtually no unders.
To overcome the 0.55 bias, I subtract 0.55 from the difference between my total and my book's total. Then I will divide this number by the standard deviation of the total population from 2002 to 2010.
With a much larger bias this season, how should I adjust? Should I subtract the difference between my total and the book's total by 1.50 instead of 0.55?
You need to find out what's causing this over bias...trying to adjust and form fit it will only make it less useful because you're trying to mimic results that you expect and not what's reality. Maybe this season has some type of variance that needs to be recognized or adjusted in your model or it's quite possible that you don't need to do anything and it's just going to be an outlier. Only you know the details of your model so it's hard to say exactly what you should do.
SBR Top-Rated Sportsbooks
NFL Model Bias
Reply With Quote