ExpertPicks College Ratings

The essence of the system (except for the first four weeks or so) is that I first do a sort routine to obtain a high degree of consistency with winners over losers. Then I impose the values of scoring differences upon this "matrix." This causes some loss of consistency, but seems necessary. There is, of course, a diminishing returns principle involved as well.

The system parameters can be changed to use more of one and less of the other, but I rarely fool with this because I simply lack time with my busy schedule at school. If I retire in the next two years (after 44 years of teaching), I may take time to research this balance more carefully. Right now it's out of the question. The predictions are dependent on this balance, so the time required to change the formulas can simply "eat you alive" if you get too involved with it.

<- Parent Directory

Herman Matthews / hmatthews@inetlmu.lmunet.edu