cHiZzLeVOL
Who Knows?
- Joined
- Apr 7, 2018
- Messages
- 19,027
- Likes
- 45,276
So, for fun (and to develop a predictive model for gambling), I ran a multiple linear regression on a bunch of variables, and the best model for predicting last year’s Sagarin rating (which factors in strength of schedule) were the 247 talent composite, the amount of returning offensive yards, and athlon’s preseason coaching rankings (after I regressed the rankings against the talent composite to strip out the effect of talent on coaching). Those three variables explained about 70% of the variability.
But...those three variables are not equal equal. By far the most important is talent, then to a lesser degree returning production, and then coaching.
Now ,that isn’t to say that I think coaching doesn’t matter. It clearly does. However, I think it’s difficult to quantify, and I think the effect is asymmetrical. My hunch is that good coaching can steal you a game or two above your talent level, while bad coaching can cause the floor to fall out well below where your talent would predict.
When 247’s talent composite comes out, it shouldnt take me long to generate a ranking for all of the teams. We should compare notes. I’d be interested to see close our results are.
I'm only going to assume you are rich, because that is a lot of time with on analysis.......