- Joined
- Oct 12, 2011
- Messages
- 12,737
Your model would essentially be establishing an upper and lower confidence level of a population.
Adjusting for the variables of any given coach is something I mentioned on Twitter last night and I agree with your premise.
Truth is, this situation really is an example of Occam’s Razor.
Just pulled up your twitter and I completely agree. The first step would probably be to adjust performance as expected against the same competition. Part of them would include creating a model that compares historical results. For example, spread variation 1 produces 37 net ppg in the FCS might only produce 30 net ppg in the FBS does to better coaches being able to better prepare players. I think a really good model would take years to make and that's with the raw data being supplied.
At this point, I am calling on Manny using any meaningful data or statistics to make decisions. Just considering the population mean of a distribution doesn't mean we fall on the right side of that. Our chances obviously change by opponent and situation. I am guessing that Manny watched Money Ball on a plane ride and instantly became a "data guy".