Every year at about this time, a bunch of college football analysts rag on the computer ratings that make up a third of the BCS ratings. Anything that disagrees with their pristine assessments of who's the best must be flawed somehow. There are some calmer voices out there, such as ESPN's Brad Edwards, but they often get drowned out by the louder ones. While the computer ratings are far from perfect, they definitely have their merit.
It's SSS, dammit! - The computer ratings to date are based on the 7 or 8 games that the teams have played so far. Naturally, strange things are going to happen when you're only using that many data points. The fact that almost half of Oregon's schedule so far has been Washington St., New Mexico and Portland St. is going to hurt them in the computers. Luckily, the ratings as they stand right now don't actually count for anything. Once everyone's played 12 or 13 games, the computer ratings should paint a much clearer and more accurate picture. Obviously, you'll still run into small sample size issues at the end of the season, but there's not really anything we can do about that until we perfect cloning.
Margin of Victory - Another reason Oregon may be lower in the computers than in the human polls is that the BCS ratings don't take margin of victory into account. This may seem counter-intuitive as using margin of victory leads to a better prediction of team success going forwards (Sagarin says this almost word for word on his site, where he has Oregon ranked #1 in his system that includes margin of victory). However, the BCS really only wanted the current iteration of the computer ratings for the purpose of a fancy strength of schedule...you'll remember that there used to be a SOS component in the BCS until 2003. Additionally, margin of victory generally plays a factor in the human polls, and the BCS wanted to avoid double counting (never mind that there are two human polls).
The nature of stats - A sentiment I've heard attributed to Bill James is that a a good statistic should 80% confirm what you know and 20% teach you something new. The reason for this is that computers can obviously keep track of more than we can. For example, most human pollsters have kept Missouri ranked much lower than the computers this year, likely because no one was talking about them before the season, and their non-conference schedule was less than sexy. However, Illinois and San Diego State have had much better years than expected, and thus Missouri has seen a bump in the computers due to those victories. I understand the importance of actually watching games and catching the subtleties that the BCS computer ratings can't catch (for example, Oregon State was a lot better when they had James Rodgers, so Boise and TCU's wins were more legitimate than they will look when OSU finishes 7-5), but we have to acknowledge that the computers simply 'know' more than we do, and can teach us a thing or two.
My Rankings:
1. Boise State
2. Oregon
3. TCU
4. Missouri
5. Auburn
6. Alabama
7. Michigan State
8. Wisconsin
9. Ohio State
10. Oklahoma
11. Florida State
12. Utah
13. LSU
14. Stanford
15. Virginia Tech
16. Nebraska
17. Arizona
18. Iowa
19. Miami (FL)
20. Arkansas
21. South Carolina
22. USC
23. Mississippi State
24. Oklahoma State
25. Baylor
Auburn at #5 probably sticks out the most. I don't think there's a great difference between the teams in the top 5, but Auburn falls at the bottom of that pile for now. They don't have the week to week dominance of Boise, Oregon and TCU, and Missouri's defense is better. Like I said though, it's close, and "Roadblock Saturday" is sure to sort a thing or two out.
Utah at #12 is a bit lower than most human pollsters as well. That's what a #113 strenght of schedule will do to you. I am impressed that they've been dominating almost everyone, but let's see what they do against the rest of their schedule first:
@ Air Force
TCU
@ Notre Dame
@ SDSU (a very good 5-2...3 point loss at Missouri, and questionable loss to BYU)
BYU
So yeah, PLENTY of room for them to move up.
Finally, #24 Oklahoma State is also ranked pretty low. Once they beat someone better than Texas A&M we can talk about moving them up. They'll get a chance this weekend against Kansas St.
This Weekend's Top 5 Games:
#5 - Baylor 24, Texas 21 - While this game feels like the end of the road for Baylor's run to the Big 12 South lead, I don't think we can count on Texas' offense to come through. Even in their big win over Nebraska, they only mustered 271 yards. Baylor's defense obviously isn't as good as Nebraska, but it probably doesn't need to be with Robert Griffin and Jay Finley on offense.
#4 - Auburn 38, Ole Miss 28 - This has been a popular upset pick this week, and while I can see it happening, I don't think Cam Newton will let it happen. The tighter the game, the more he's just kept the ball himself and dominated. Last week's 200+ yards against LSU shows what he can do against a good defense; against Ole Miss it could be even more ridiculous. Yes, Auburn's defense will probably be far from perfect, but I don't think it matters.
#3 - Oregon 41, USC 31 - You can basically copy the Auburn-Ole Miss write-up into this area. Yes, Oregon's lack of defense will probably catch up to them at some point, but I don't think it happens here....too much motivation against USC.
#2 - Missouri 31, Nebraska 23 - I want to pick my Cornhuskers, but I just can't. Missouri matches up well with Nebraska's defense. Nebraska's main strength is causing incompletions (Locker and Gilbert had 25% passing days against UNL), and Missouri's short passing game combined with their ability to run efficiently can effectively neutralize that.
#1 - Iowa 30, Michigan State 21 - The main difference in the seasons of Michigan State and Iowa has been special teams play. Iowa's meltdown in Tucson and missed XP last week were main culprits in their losses, while Michigan State's fake punts have saved them in tight games. However, I think Iowa is the better team on offense and defense, and Michigan State has to be running out of fake punts.
