I am growling getting frustrated with some of the analysis on the BlogPoll. Some of the discussions I had in the comments of the draft ballot really got me thinking.... A lot of the philosophy behind the blogpoll is very flawed.
If you go back to the beginning, the idea of the BlogPoll was to figure out who would beat who. The idea was to power rank the teams. Even though I wasn't a voter from the beginning, I came across this article when I first started to do this, but a lot of Brian Cook's analysis was purely based on what just happened like "despite team x losing this week, they are still on the ballot". In a later post, he talks about the BlogPoll being based ONLY on resume now.
Well, that doesn't make a lot of sense either, based on his analysis, because he isn't really looking into the resume of teams. For example, last week he thought Missouri was too high on some people's ballots. The thing is, I thought it was too high also, because I was looking at the predictors, which had them in the teens. The resume based rankings had Missouri around 10th/11th across the board, however. So what resume is he looking at? Who knows?
Below is a resume based approach on ranking the teams. I used Sagarin's ELO_CHESS for simplicity, as it factors SOS and looks at Wins and Losses only. The ballot is below. My initial guesses is that Mr. Cook is going to think its silly to have Oregon State on the ballot, but why? The have the #1 SOS according to Sagarin, and they have lost only 1 more game than a two-loss Texas team with a SOS in the 40s, that will likely be on the ballot. He will probably also think having Air Force receiving votes as being silly, as they just lost to SDSU. Something is to be said about Michigan State too. Does anyone buy they are the 3rd best team? Well, too bad! The resume says they are!
But that's the whole thing. You can't have it both ways, and often times in the analysis, he wants it both ways. I also think it's incredibly stupid to have a poll based on resume-only, as any one of hundreds of computer models out there already do this. The idea of creating a power ranking where you use some analytical data (predictive models), and your own opinions about football (in terms of who would beat who) seem to make a lot more sense.
It will be interesting to see how the poll shakes down, and the analysis that is written about the ballot this week. Please accept my apology in advance for throwing out an approach I tried to take. I am just experimenting to see if I do it the way it's written (at least according to the most recent update to the philosophy), if Mr. Cook has any issues with the ballot. He probably won't even notice though...
Clone Chronicles Ballot - Week 8
|3||Michigan St. Spartans||21|
|6||Boise St. Broncos||-4|
|7||TCU Horned Frogs||-4|
|11||Florida St. Seminoles||1|
|14||Ohio St. Buckeyes||-7|
|18||Oregon St. Beavers||1|
|19||Oklahoma St. Cowboys||3|
|20||Alabama Crimson Tide||-9|
|21||Kansas St. Wildcats||--|
|22||Mississippi St. Bulldogs||--|
|23||Nevada Wolf Pack||-5|
|25||Air Force Falcons||--|
|Dropouts: South Carolina Gamecocks, Arkansas Razorbacks, Virginia Tech Hokies, Florida Gators, Miami Hurricanes, California Golden Bears|
SB Nation BlogPoll College Football Top 25 Rankings "