On Nov 17, 2008, at 12:51 PM, [EMAIL PROTECTED] wrote: > So the formula looks like this: 1-((computerrank-1)/25)=BCS Score > > Right now the Gators are ranked 5-5-5-2-6-6. You drop one of the > 6s and the 2 and average the remaining 4 = 5.25 or a BCS rank of .83 > Texas is ranked 3-3-3-1-2-3. They drop a 1 and a 3 and average out > to 2.75 or a BCS rank of .93 > > If the two polls that have us ranked 6th (interestingly enough, > they both start with the same data source) move us up to 5th, our > BCS rank improves to .84, right now, we would need to be at a > computer rank of .88 or ranked 4th (in the computer polls) to > overtake Texas. By the time we beat bama, we will have moved to at > least 4th in the polls, and probably 1 or 2 in some of them. So > don't worry, we win out, we will be in.
O.K. I guess that my point is the conversion from the raw score to the rank is a problem as an artificial gap is created for teams. It appears that it is only done for the computer polls and not the two human polls. Two teams can be exceptionally close in their raw score, but once converted to rank, a larger gap can be introduced (at the lower levels, I suppose the opposite effect will occur as well). Picture this. If the computers were in complete agreement, there would be an even 0.40 difference in their computer component (the #1 team would have a 1.00, the #2 would be 0.96, the #3 would be 0.92, etc.). However, the human polls keep the raw score and do not convert to rankings. So, the spread between two teams next to each other in ranking can be very small (in the USA Today scoring, #12 Okahoma State is 0.0039 ahead of #13 Georgia), or very large (#13 Georgia is 0.0931 ahead of #14 BYU), a gap almost 24 times as large. This inconsistency in how the scores are equalized can result in the computer poll having a greater influence than the sum in the human polls, belying the claim that the human polls are "2/3rds" of the scoring. For example, #4 Texas is 0.0112 ahead of #5 Oklahoma in the USA Today scoring. If there were a similar gap in the Harris poll, then the standard 0.40 spread in the computer rankings would determine which was #4 and which was #5, regardless of whether the human polls were in agreement. -Zeb --~--~---------~--~----~------------~-------~--~----~ GATORS: ONE VOICE ON SATURDAY - NO VOICE ON SUNDAY! 1996 National Football Champions | 2006 National Basketball Champions 2006 National Football Champions | 2007 National Basketball Champions Three Heisman Trophy winners: Steve Spurrier (1966), Danny Wuerffel (1996), Tim Tebow (2007) - Visit our website at www.gatornet.us -~----------~----~----~----~------~----~------~--~---

