The Coaches Poll -- Comparing Start to Finish

Friday's release of the preseason coaches poll got me thinking. When I saw the Virginia Tech Hokies sitting there at #6, I immediately thought, "man, yet again, Virginia Tech is over-rated." But then I thought rather than rely on my own recollections and biases, I should see if my reaction was true.... that Virginia Tech does, in fact, start the year ranked higher than where they finish. As you will see, I couldn't have been more wrong on the inaccuracy of preseason voters when it comes to VT.

I began by collecting the coaches poll rankings as far back as readily available (ESPN's complete poll archive goes back to 2003), pulling in the preseason rankings and the final (post-bowl) rankings. I have seen some lazier analysis that used average rank; this, in my view, is not appropriate. I instead use the "points" awarded on a ballot, which allows us to measure stronger rankings (a consensus #1 is stronger than a divided #1). Points also allow us to measure those teams that fall into the "others receiving votes" category. Teams that don't show up in a poll at all get a big fat zero (which I think is fair for this exercise), and their average poll position/points will suffer as a result. I also divided the points by the total number of voters, to eliminate the fluctuation in how many coaches completed ballots. Basically, if a team ranked a perfect #1 on every single ballot, they get 25 points, and if they ranked a perfect #25 on every ballot, they get 1 point.

We begin by looking at the best finishing teams from the 2003 to 2009 seasons. Southern Cal truly dominated this time series, and edged out Texas for the #1 spot. USC's 20.6 "points" means that on every final ballot from 2003-2009, they had an average of 20.6 points, which means they were, on average, ranked about fifth (26 minus 20.6 equals 5.4). Ohio State, LSU and OU round out the top 5. In the top twenty, there are five SEC teams, five Big 10 teams, two Big XII teams, two Pac 10 teams, two Big East teams, and four from other less powerful conferences.

To help frame the discussion on which teams are beating or failing to meet expectations, we also need to see how the typical preseason rankings look. We see a lot of similar names on this preseason list when compared to the final rankings, but some names such as Notre Dame, Louisville, California, Florida State and Tennessee also show up, suggesting these teams fail to hold onto loftier preseason rankings in the final rankings.

With these two lists compiled, we can take the next step to figure out which teams appear to be perpetually overrated in the preseason. Basically, we can compare the average final ranking against the average preseason ranking to arrive at a list of teams that tend to be the most overrated. Oklahoma tops this list, averaging almost 7 ranks below their preseason ranking. Florida State, Michigan, Miami and Cal round out the top five "disappointers." All of these disappointing teams come from so-called "power conferences."

Looking at the reverse phenomenon, the best "outperformer" compared to its typical preseason rank is Boise State, which tends to climb about 9 spots in the polls during the typical season. TCU, Utah, Alabama and Cincinnati round out the top 5. It will be interesting to see how Boise fares this year, as they have usually been a teens/twenties ranked team in the preseason polls, but the pollsters seem to have finally given them some respect to start off the season.

And yes, Virginia Tech ranked with an average of 12.3 points in the preseason, and an average of 12.3 points in the final poll. I guess the preseason polls couldn't have had them more right.