Creating a power numberpreseason for each team in college football and basketball?
I have struggled the last 2 season in ncaaf creating a number for smaller (mac and sunbelt) conference under the radar teams.
I use 6 variables
Conference grade
coaches
experience of players and success in past
home field vs away splits
programs expectations and motivation
national exposure
key losses to graduation and nfl draft
My biggest issue has become finding enough information and how to grade a coach with none or close to no experience in the past.
Anyone who could chime in with a helping hand would be greatly appreciated and of course Ill give points etc
Thanks, MILK
Yeah, smaller teams are tougher. For the most part, the variation in quality from the major schools is quite small year to year, and thus the past year(s) is(are) a pretty good indication of the starting strength for this year. Most of these schools have lots of depth at almost any position and even injuries or players graduating have very little effect.
Smaller teams and smaller conferences are much tougher. One thing to consider would be a year-to-year roster differential, particularly for starters and possibly 2nd stringers. Roster changes could then be weighted against the previous year's performance in a given metric/set of metrics. For example, if a good passing team loses its starting quarterback, you may want to substitute conference average passing stats in for your ratings. If, however, an average team lost their QB, you may not adjust the rating much.
Are you doing this by hand, or do you have a model you're feeding this into?