Quote:
Originally Posted by chuckleslovakian
I guess it is a flaw of computer analysis. As long as you solidly beat Samford, and don't get anyone injured I consider that a win. Sure the computer said you could of beat Samford by more, but no reason to keep running Chubb against Samford.
Yeah, I posted about that issue with achievement rankings last week. I'm not sure exactly the best way to handle it, in part because "achievement" is not so simple to define.
On one hand GA only beat Samford by like 28, and Bama would be like a 50 point favorite over Samford. On the other hand a win is a win. On the other hand "only" winning by 28 is still kinda meh. Like rolling over a team should get you up by 40-50 before you even can realistically get your second string in.
I'm starting from a similar mathematical place that goes into power rankings. Those work well, I've had a few years with them, they have a clear objective (pick winners). They seem to handle the "didn't beat a cupcake by as much as expected" case really well - adjusting a team's rating down but not by a ton.
So on one hand starting with that framework for achievement seems good, on the other hand who knows.