re: BCS works __% of the time? 27%

quote:

---

I see your point but it's a flawed formula. You can't just take the records without regard to which conference the team plays in...ex I would take Oregon St over Wyoming, N Texas and maybe even Pitt in this example. Just looking at it, basically every decent team NT played, beat them. You wanna know who Toledo beat that year? Eastern Mich, Temple, Ball St, West Mich, Ohio, Central Mich, No Ill, Bowl Grn, and MiaO. Take another hypothetical...say team A plays and beats the respective champs of the Sun Belt, MAC and Conference USA, all of whom end up with a 10-2 record. Is that better than team B beating a 7-4 Big XII team, a 8-3 Pac 10 team and a 7-4 ACC team? I say no.

---

1. I have no idea how the models ranked each one of these teams relative to each other and I'm not going to look it up. But I'm extremely confident that beating more teams with winning records and losing to fewer teams with losing records is going to be looked on favorably.

2. The one thing you do not want to do with a model is to base it on subjective beliefs of who is better than who. That should be the output not the input. If you were to look up how these teams were ranked, the models may either confirm or contradict some of your beliefs. You don't have to accept the model's results but it is a good thing to have preconceived notions challenged.

3. Subjective judgements are necessary and these are the realm of the human polls.