No wonder his computer rankings are so worthless that they are tossed out over half the time in the BCS rankings. From ESPN's "I-Formation":
I don't understand the logic behind this. If you beat a team in October that ended the year ranked #3, how is that easier than beating that same team in November? I understand how a team's derived ranking early in the year is not as meaningful as their derived ranking at the end of the year, but don't ALL of the computer rankings take into account what a team has done up to date, not just what it had done prior to an individual game?
But wouldn't that still be based on the final ranking? If OU beats a Tech team and then moves up in the rankings then we would have beaten on the final rankings a top three team even though we did in Oct?????
The article also mentions Texas's weak non-conference sched, and the best team we beat was Rice. Who cares about the fact we beat Mizzou, and Oklahoma doesn't play them. I'm willing to be Mizzous is better than TCU and Cincy.
Most of the computer rankings have severe flaws. Many were made by sports people who are pretending to be mathematicians. Their methods would make anyone who works in predictive modeling cringe. Also, the BCS should NOT use any closed models. They should be open to public review and scrutiny. Billingsley is worse than most because he models the polls, not the teams.
Billingsley is assuming that rankings earlier on in the year are less precise and therefore discounts them. Specifically, he assumes that higher ranked teams earlier in the year may not be ranked that high later in the year. Therefore, beating number one early on is not as good as beating number one later.Um... OK. That team never should have been ranked that high. Is that what he's saying? Well, what about a Florida? Who lost to Ole Miss. At home. Explain that, Mr. Math. Is it better to for Florida to show the world their tighty-whities in the girls gym, or in the middle of the cafeteria during lunch?
