McGeesPub
Well-Known Member
- Joined
- Dec 22, 2011
- Messages
- 4,915
- Likes
- 4,745
I've been poo-pood on by much better, to be sure.
The problem is that when you look at these numbers over a long enough timeline you begin to see very clearly that what most people consider to be determining factors in football, simply aren't. Using only these numbers you can arrive at a 70% prediction rate. That means that literally every other variable you can imagine falls within the 30% of the time the numbers don't work.
coaching.
attrition.
weather.
home-field advantage (very real).
Gruden's tie color.
Those all could arguably account for the 30% of the time the evaluations don't work, and I get that. But, talent averages alone can account for the other 70%.
EDIT: since 2005 (the farthest back this data goes) there has only been 1 time in a national championship game that a team with a worse talent average has won the game. Talent is almost a 100% predictor in national championship games.
Subtle dig aside, I'm curious, do these projections take into account a class rank at the time of the signing day, an adjusted class rank to account for transfers, etc, or each individual player's recruiting ranking at the time the model is run?