Bumped. Tasser offers fodder for some good discussion. GO BRUINS. -N

I came across this interesting web site yesterday, which was mentioned in Stewart Mandel's college football mailbag.  The author has basically gathered all preseason rankings from various magazines since 1993, and compared them to the final rankings for each of those years.  I found it very interesting, so I made a spreadsheet (slow summer work day...) in order to see if there has been any evidence of bias in these rankings. 

The way the analysis was done on the web site was to use the difference between a preseason ranking and the final AP ranking.  A negative difference means overrating (finishing with a lower ranking), while a positive difference means underrating (finishing with a higher ranking).  All unranked teams in the postseason and final poll were given a ranking of 26.  While it was difficult to discern and overwhelming trends, I did notice the following:

• Compared to the predictions for the other BCS conferences, the predictions for the Pac-10 have been wildly off the mark.  On the other hand, those for the SEC and Big 12 have been comparatively accurate (can you say top heavy?).  I can think of two reasons for this:  (i) either the prognosticators spend little time researching/viewing teams from the Pac-10, or (ii) the Pac-10 is one of the most competitive conferences with the most parity.  I tried to see which magazine predicted the Pac-10 best, but they have all been way off at some point, no single magazine stood out.
  
• Under/Over-rated: It would take too much time to check every school, but of those I looked at, Oregon (total of +56) and Wazzu (+59) have been consistently and significantly underrated, while Michigan (-73), Tennessee (-72), Florida State (-71), Texas (-57) and -surprise- Notre Dame (-57) have been mostly overrated.  UCLA ends up with a total of -12, and U$C gets -32.  This isn't exactly accurate, since a single year of major overachieving or underachieving can skew the results, but you get the general idea.
  
• Florida, Florida State, Michigan and Tennessee are the only teams to be ranked in the preseason every year since 1993.  No team has finished ranked every year since 1993.  Georgia has the longest current streak, being ranked since 1997.
  
• Predictions on UCLA have been wrong nearly every time.  UCLA was ranked in the preseason six times since 1993, and of these six times it finished ranked in the top 25 ONLY ONCE!  We finished with a top-25 ranking four times since 1993, of which three times was after not being ranked in the preseason.  The only year in which they were right was 1998, where we were ranked 9th in the preseason and 8th in the final AP poll.
  
• U$C sucks.  Oh sorry, we're talking about rankings.  From 1993 to 2000, the trOJies were significantly overrated.  They were underrated in 2002 and 2003.  Since then, they've performed almost exactly as predicted.  I predict that Pete Carroll has three nipples (but hey, I might be wrong).

Many more fun things to check out for stat geeks, and if anyone is thirsting for it I can send them my spreadsheet. 

Let's hope they're wrong about UCLA again!

Go Bruins!