The model can not fully predict performance where the opposition has had a 20-30% turnover in personnel while the Goph's have the same turnover. And, the supporting cast of 2 and 3 deeps is now very different. So, the whole model becomes even more tenuous. Not to speak of injuries, odd football hops, weird ref calls, and hangovers, girl friends, and academic load, and the odd case of food poisoning. It doesn't figure in more goodness of fit of players to system, conditioning, staff insight, or year 3 of the Kill regime. I'm leaning on the intangibles playing a larger than life role than last years or last 30 years of statistics can prove.
Could the jump in defense last years defensive performance be predicted from the prior 1, 2, 3 or 4 years of data? I doubt it. See the intangibles.
Were the number of offensive linemen injuries predictable from the prior 1, 2, 3, or 40 years of data? Nuh uhh.
These are nice slices of data which can be pointed toward for areas of improvement, but then again, the better numbers won't be found in those very broadly based numbers. To be a good data driven decision maker requires having the right numbers to begin with. This piece was a pure fail.
