By week 6, everyone has hit conference play, so the competition gets tougher and games become harder to predict, as evidenced by the huge drop-off in accuracy this week. The upsets start to mount as the stress of the season wears on and teams start slipping due to injury. Expect percentages to hover in the middle to low 70s for the remainder of the season.
Eddie
Accuracy: 39-16 (70.9%)
The Good: Overall, this win percentage was very good relative to other computer algorithms, which also had difficulties predicting the upsets. Among the 58 prediction algorithms at PredictionTracker, RBA finished 5th in week 6.
The Bad: Missing 16 predictions implies that the algorithm needs a separate system to analyze conference games versus non-conference games.
The Ugly: Toss-up between Penn State-Illinois and Miami-Florida State. Penn State has had an atrocious offense all year, but they remain highly ranked. On the other hand, Florida State has actually looked pretty good with the exception of the Oklahoma game and are likely underrated by RBA.
Justin
Accuracy: 40-15 (72.7%)
The Good: Finally starting to make up some ground on Eddie, and tied for second place for the week on the prediction tracker.
The Bad: Despite an up week, I'm still a bit disappointed in how I've done overall. I suppose getting nearly 78% correct for the year is nothing to really laugh at, but I just feel there's a lot more that can be done to improve the system. I have some ideas, but just lack the time to implement them, which is all the more frustrating. I disagree with Eddie, though, that a separate system is necessary to handle conference play.
The Ugly: I'm not sure if it's being 85.3% certain that Penn State was going to beat Illinois (whoops) or 85.9% certain that Alabama was going to beat South Carolina. To be fair, rarely has one team seen such a tremendous drop on one side of the ball. At this point last year Penn State had an offensive efficiency of 23.6 PPH, compared to 16.7 PPH this year. To put that in perspective Penn State's defense is the equivalent of Nebraska's but their offense is struggling to stay ahead of Iowa State. To say the bottom has dropped out would be unfair to things for which the bottom has actually dropped out.
Overall Season Totals
Eddie: 229-61 (79.0%)
Justin: 226-64 (77.9%)
These predictions currently put us ahead of two of the BCS computers (Billingsley and Sagarin). Eddie is only one game behind another (Massey).
