| Rank | +/- | Team | WinPct | SoS | Adjusted | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Off. | Def. | Pace | |||||||||
| 1 | -- | Alabama | 1.000 | 0.550 | 12 | 25.8 | 6 | 4.8 | 1 | 157.1 | 120 |
| 2 | -- | Stanford | 0.992 | 0.542 | 27 | 28.8 | 2 | 10.1 | 8 | 167.1 | 52 |
| 3 | -- | LSU | 0.992 | 0.551 | 11 | 25.0 | 8 | 7.2 | 2 | 160.5 | 112 |
| 4 | -- | Oregon | 0.974 | 0.530 | 45 | 26.0 | 4 | 9.2 | 4 | 176.1 | 3 |
| 5 | -- | Oklahoma | 0.974 | 0.545 | 23 | 25.3 | 7 | 9.5 | 5 | 171.4 | 15 |
| 6 | +2 | Boise St. | 0.966 | 0.461 | 94 | 25.9 | 5 | 10.2 | 10 | 168.9 | 32 |
| 7 | -1 | Oklahoma St. | 0.957 | 0.549 | 16 | 27.4 | 3 | 14.7 | 50 | 172.9 | 8 |
| 8 | -1 | Wisconsin | 0.949 | 0.536 | 39 | 28.9 | 1 | 11.2 | 14 | 162.0 | 107 |
| 9 | -- | Southern Miss. | 0.941 | 0.484 | 75 | 22.0 | 14 | 14.7 | 49 | 168.6 | 36 |
| 10 | +1 | Clemson | 0.920 | 0.526 | 57 | 20.4 | 25 | 13.4 | 36 | 166.6 | 63 |
| 11 | +5 | Florida St. | 0.920 | 0.550 | 13 | 21.0 | 20 | 10.1 | 9 | 166.0 | 72 |
| 12 | +1 | Ohio St. | 0.919 | 0.529 | 49 | 19.1 | 37 | 10.7 | 11 | 159.2 | 117 |
| 13 | +2 | Arizona St. | 0.912 | 0.529 | 51 | 20.2 | 27 | 12.0 | 20 | 173.9 | 6 |
| 14 | -4 | South Carolina | 0.911 | 0.543 | 26 | 20.5 | 23 | 9.8 | 6 | 158.0 | 119 |
| 15 | +4 | Arkansas | 0.910 | 0.567 | 1 | 22.7 | 13 | 12.7 | 29 | 165.6 | 78 |
| 16 | +7 | Notre Dame | 0.902 | 0.564 | 2 | 20.1 | 30 | 11.8 | 19 | 164.8 | 86 |
| 17 | -3 | TCU | 0.893 | 0.461 | 93 | 23.1 | 10 | 13.1 | 33 | 168.0 | 42 |
| 18 | -- | Nebraska | 0.893 | 0.530 | 47 | 20.6 | 22 | 13.0 | 32 | 166.8 | 60 |
| 19 | -7 | Georgia | 0.874 | 0.550 | 14 | 22.8 | 12 | 12.3 | 24 | 159.7 | 114 |
| 20 | +2 | USC | 0.857 | 0.534 | 41 | 20.0 | 32 | 12.3 | 25 | 166.9 | 58 |
| 21 | -4 | Florida | 0.853 | 0.555 | 8 | 20.8 | 21 | 8.3 | 3 | 161.5 | 108 |
| 22 | -1 | Michigan St. | 0.850 | 0.543 | 25 | 18.5 | 42 | 11.7 | 18 | 167.1 | 54 |
| 23 | +1 | Iowa | 0.838 | 0.522 | 60 | 18.3 | 43 | 13.4 | 38 | 163.8 | 93 |
| 24 | -4 | Texas A&M | 0.829 | 0.554 | 9 | 20.5 | 24 | 12.1 | 21 | 172.4 | 11 |
| 25 | NA | Texas | 0.805 | 0.523 | 59 | 21.1 | 17 | 11.1 | 13 | 168.2 | 40 |
New entries: Texas.
Dropped out: Virginia Tech.
You idiot! LSU beat Alabama last weekend! This is entirely true. Our systems have always been about predicting what happens tomorrow, not what happened yesterday. Here's what I saw on Saturday night -- four missed field goals and a fluky interception at the LSU 1. RBA says that if they played again next weekend on a neutral field that (1) Alabama would beat (3) LSU. I'm completely comfortable with that. Does that mean that Alabama should be ranked ahead of LSU in the BCS? No, because they are retrograde rankings, not predictive rankings.
You idiot! Southern Miss isn't a top 10 team! Generally, I'm tempted to believe you on that. Then, I went back and looked at how they got here. Last weekend, the (9) Golden Eagles went on the road and scored 48 points by a 97-yard interception return, a 79-yard interception return, a 60-yard punt return, a blocked punt return, and a few conventional scores. The week before, they hung 450 yards and 31 points on UTEP. The week before that, their defense held SMU to a mere field goal, as the offense rang up another 453 yards and 27 points. The Golden Eagles plastered Navy, 63-35, with 584 yards of offense. They scored 48 on Rice with 654 yards of offense. The only close game of the past two months was at Virginia, where the Golden Eagles managed only 374 yards in a 30-24 victory. Does that mean that Southern Mississippi is going to crash the BCS party? No, because they haven't beaten anybody. However, I wouldn't want to play them right now.
You lazy bum! I thought you were going to fix this! I looked into the Working-Hotelling Method last week. The basic idea behind the Working-Hotelling Method is that linear regression has higher uncertainty in areas of the graph where there are few points. Southern Miss is elevated to a top 10 team because they have not played enough games against strong competition to see the natural degradation of offensive efficiency that usually occurs as competition becomes more difficult. As I looked into the Working-Hotelling Method, I came to realize that it only affects the confidence intervals, not the actual prediction itself. I could theoretically include the uncertainty as a negative modifier in the predictions, but that requires a fairly substantial modification to my algorithm, and I'm not terribly sure if its mathematically rigorous. I'm ordering a pricey book on linear regression, so we may see the modifications in time for next year's picks. In the meantime, take heart in knowing that the Golden Eagles' offensive slope is dropping.
Follow us on Twitter @TFGridiron
