On Pre-season Polls And Their Predictive Powers

Pre-season polls are everyone's favorite. Just like sitting next to a TCU fan on an airplane is everyone's favorite. These semi-meaningless rankings of college football teams never fail to incite debate (of both the civil and non-civil variety) and generate clicks for websites. Long have these polls been scoffed at for their abhorrent inaccuracies and their pumping up of certain neighbors to the Southeast. But are these claims valid? How accurately do these polls predict the final poll? Let's find out.

Methods

I took the AP pre-season polls from 2011 to 2014 and compared them with their respective final polls. Each team in the preseason poll was compared to where they finished in the final poll and a "delta ranking" was obtained. Some teams began in the Top 25 and then finished without a vote. For these teams, I assigned them a ranking of 50 for the final (the AP poll usually had 15-20 teams "receiving votes" in the final poll, so I figured 50 was a nice number to use).

2011

The first thing one notices is that Oklahoma was highly over-ranked (this is a recurring theme; they finished -11.75 games below their initial rankings over these four years). Florida State, Texas A&M, Notre Dame, and Ohio State also had pretty significant drops. USC was the highest mover of the year, with a climb of 19 spots. Overall, the average delta was -6.08, the best of the four years. One team from the 25 finished without a vote and 14 teams dropped from their preseason ranking (56%, about what one would expect). 2011 was pretty average as a whole.

2012

Ohhhhh USC. The preseason number one tumbled out of the rankings and finished the season without a single vote. Bama, being Bama, won the championship again. LSU and Oklahoma continued their over-rated ways, dropping eleven spots each. There were five teams that finished the season without a single vote, a good 20% of the poll: USC, Arkansas, West Virginia, Michigan State, and Virginia Tech. With an average delta of -8.72, this year was pretty bad for the pollsters.

2013

Florida won the title of biggest loser in 2013, dropping 40 places in the polls (honestly, it was probably more than that). Michigan was not far behind, falling 33 spots. This year was also the only year of the four in which a team came from outside the Top Ten to place number one (Florida State). Auburn, who finished second that year, didn't make the pre-season poll, nor did third-place Michigan State. With six teams failing to receive a vote, this year was clearly the worst year for pollsters. The average delta was -11.08, a huge gap. Oklahoma was under-rated, for once, though. Texas was highly over-rated, as we all knew going into the season. Nearly 75% (18/25) of the pre-season poll finished below their initial ranking. 2013 was a mess.

2014

The second-best year for the AP out of the four. Pre-season number one FSU went undefeated in the regular season to make the CFB Playoffs (much to the chagrin of BU and TCU fans). Ohio State started out at five and ended up at one and Alabama and Oregon stayed pretty constant as well. Then there is Oklahoma, South Carolina, and Stanford. All three dropped over 30 spots. The aggys shellacking of South Carolina early on brought them much attention in the rankings, but they ended up barely making the "received votes" section of the poll. Typical aggy. Anyway, three teams failed to receive a vote at the end of the season, 14/25 fell in ranking, and the average delta ranking was -7.96. All in all, a pretty decent year for the AP, with some major exceptions in Oklahoma, etc.

What does this mean?

Is the SEC over-rated? Yes, of course they are. But do the numbers back up that opinion? Yes, sort of. Quantifying "over-ratedness" is somewhat difficult given the manner in which I have presented this information. In terms of teams that were in the Top 25 for the pre-season poll, yes, the SEC is over-ranked. In fact, every league is:

The SEC teams in the pre-season polls finish, on average, -9.5 games below where they began. The Big 12, it seems, is the most over-rated; perhaps a product of Oklahoma and Texas being ranked highly and then failing or that Baylor, who won the Big 12 in 2013 and 2014 (#B2B), did not even begin in the polls in 2013 (They were 27. If you include their climb to 13th in the average, the Big 12 comes just behind the B1G). The most surprising result is the ACC. Yes, they are not really as good of a league as the other four, and yes, FSU has a really easy schedule, but it appears the for the few teams that do get ranked (FSU, Clemson, and Louisville), they tend to maintain their rankings pretty well.

Although this method of ranking is slightly biased in favor of the negative outcome (a good team that started out highly ranked can only go up so many ranks, whereas a team that starts out high can drop heavily), I think it is still a pretty good representation of what pre-season polls are: clickbait. No one can predict the future, so take those pre-season rankings with a grain of salt, rub that salt in a TCU fan's open wound from October 11th, and get ready to unleash LaQuan McGowan.