How accurate are the preseason polls? Here's the facts...

Nov 29, 2015, 4:12 PM

Preseason vs Week 14 AP Rankings.pdf(44.6 K)

Just for fun, I decided to look a the AP preseason rankings vs the AP week 14 rankings to see how accurate the AP preseason polls fare to the reality of field play (see attached). The AP overrated 5 SEC teams (Auburn, Georgia, Arkansas, Missouri, Tennessee) that are no longer in the top 25, overrated 3 PAC-12 teams (UCLA, Arizona State, Arizona) that are no longer in the top 25, 1 ACC team (Georgia Tech) that is no longer in the top 25, and 1 Mountain West team (Boise State) that is no longer in the top 25.

On a comparison of the AP week 14 to preseason top 25 rankings they underrated 3 Big 10 teams (Iowa, Northwestern, Michigan) that were unranked in the preseason poll, 3 AAC teams (Houston, Temple, Navy) that were unranked in the preseason poll, 1 ACC team (North Carolina) that was unranked in the preseason poll, 1 SEC team (Florida) that was unranked in the preseason poll, and 1 PAC-12 team (Utah) that was unranked in the preseason poll.

If you give the AP credit for picking the final regular season top 25 teams in the preseason poll (regardless of their top 25 rank order) the AP preseason poll correctly predicted 15 of the top 25 teams for a 60% accuracy rate. Another way of looking at is that 40% of the ranked teams in the preseason AP top 25 are now unranked at the end of the regular season.

So what conclusions can be drawn from this?
1. The SEC and PAC-12 were way overrated in the early polls. The Big 10 and AAC were the most underrated conferences. An early SEC poll bias does exist and could be resulting in unfair rankings advantages to certain SEC teams as the season progresses.

2. Preseason polls are largely inaccurate (60% accuracy) and should be taken with a grain of salt. My personal opinion is that no polls should be released until after week 4 of the season to avoid the current conference bias that clearly exists in polling.