Using the right season's play data made a big difference this week, as the model correctly predicted 74.1% of games correctly. That was almost exactly the overall season accuracy of 74.0%.
There were a few big changes in expected win totals from before Week 7 to before Week 8. In the Big 12 Texas, based on its convincing win over Oklahoma, jumped 13% from 5.91 to 6.68 wins. In the Big Ten, Nebraska turned it's mangling of Purdue into an 8% jump from 7.97 to 8.63 wins. Northwestern was the biggest percentage loser in the Big Ten, falling 10% from 7.13 to 6.44 wins. In the Pac-12 USC's surprise victory over Arizona into a 22% gain in expected wins, jumping from 6.77 to 8.24 wins. Arizona felt the sting of that upset, falling 10% from 7.70 to 6.92 expected wins. In the SEC, Auburn used it's not-particularly-surprising-lopsided victory over Western Carolina to climb 24% from 6.99 to 8.67 wins. Florida's 6-17 loss to LSU, combined with Missouri's strong performance led to a 12% drop from 8.10 to 7.16 expected wins.
|TeamA||TeamB||Team 1 %||Team 2 %|
|Miami (Florida)||North Carolina||0.899||0.101|
|Ball State||Western Michigan||0.883||0.117|
|Georgia State||Texas State||0.258||0.742|
|Kent State||South Alabama||0.499||0.501|
|North Texas||Louisiana Tech||0.626||0.374|
|Northern Illinois||Central Michigan||0.733||0.267|
|Rice||New Mexico State||0.914||0.086|
|Southern Mississippi||East Carolina||0.248||0.752|
|Texas Tech||West Virginia||0.808||0.192|
|Utah State||New Mexico||0.709||0.291|