Monday, October 4, 2010
Conference Power Rankings: Week 5
By Albert Larcada
Not much movement from last week as most schools have started into their intra-conference schedules.
The Big Ten did flip-flop with the Pac-10 for the number two slot, although little change occurred in their actual ratings. Barring a catastrophe from one of their top schools, expect these two conferences to be back and forth for the foreseeable future. They seem to have a fairly firm grasp on the two and three positions in our rankings.
For the first time all season the SEC has lost it's number one computer rating to the Big 12. Computer ratings are generally a good indicator on the depth of a conference, and the Big 12 has only one member (Kansas) currently outside the top half of the 120 FBS schools in computer rating average. No other conference can make that claim.
Nevada is helping the WAC stay above the improving ACC with their ascension up the AP poll (currently 21st, 286 points).
A couple minor housekeeping notes on the system we have created. As we mentioned in our Week 3 ratings, we started to use the AP poll in our calculations instead of the Coaches poll. The only reason we made this switch was because of the Coaches poll ban on voting for USC. Continuing with the Coaches poll would have unfairly rated the Pac-10 with one of their tougher schools removed.
Also, a quick refresher on the intention of these ratings and what it actually calculates:
We created this system to grade conferences based equally on two key components, the quality of the upper tier schools AND the depth of the conference. The AP (or Coaches) poll is a good proxy for the strength of the top schools in each conference. On any given week there are roughly 35-45 schools given a positive voting number, that is, at least one 25th place vote. This means about two-thirds of FBS schools are treated equally and given zero AP points. So, only conferences with these elite schools will get any points at all and only the elite of the elite will receive the major points given out if you can make the top ten.
The second aspect of this system is designed to proxy the depth of the conference. Computer ratings are great for this purpose as they always will rate at least all FBS schools. Thus if you are a very top heavy conference with bad schools near the bottom of your standings, computer ratings will heavily penalize you.
A simple statistical process called "normalization" is then used to put the AP poll and the computer rating averages on the same scale with their 50/50 weights.
Hopefully this sheds a little more light on our methodology. Suggestions are always welcome and debate always encouraged.