Preseason Polls Are Like…

For UGA, and others, have preseason polls been a good indicator of where teams will rank in the end?

By Patrick Garbin—Twitter @PatrickGarbin

You’ve probably heard similar analogies before… Preseason polls are like a rocking chair: it’s something to do, but it doesn’t get you anywhere; a blind date: totally misleading and pointless; a beauty pageant: where you’re drawn to the hot girls, but have no idea whether or not their heads are only filled with hot air.

As the college football annuals are beginning to hit newsstands, revealed are plenty of seemingly insignificant preseason rankings which, on the whole, will appear very similar to the most preeminent preseason poll of them all—the Associated Press’—when unveiled approximately three months from now. Nevertheless, as Iowa head coach Kirk Ferentz stated about a year ago: “Preseason polls are like recruiting class rankings; it doesn't matter until you start doing something."

Au contraire, discovered a few months ago, recruiting class rankings—at least, those compiled by Rivals—do matter, and before somebody starts “doing something”: Yielding a correlation coefficient (a number between −1 and +1 measuring the strength of the association/relationship between two variables) of .793, there is a strong, and nearly very strong (.8 to 1.0), correlation between Rivals’ team recruiting rankings and the final AP rankings over the last 15 years.

Accordingly, I decided to see how strong of a positive relationship—if positive at all—there has been between the preseason AP poll and the final AP rankings in regard to Georgia, and several neighboring schools of interest. Beginning in 1950, the initial year of the preseason AP poll, through 2016, the preseason and final rankings of Georgia and eight other often nationally-ranked teams from the South were correlated annually to reveal their correlation coefficient, or r. Simply, think of "r" as how efficient the Associated Press has been at preseason ranking each team in association to where it finishes in the AP's final poll (whereby if r is between 0 and .200, there is a very weak positive relationship between the AP’s preseason and final polls; .200 and .400 is weak; .400 and .600 is moderate; .600 and .800 is strong; and .800 and 1 is very strong).

Ranked according to r, each team is also listed with the number of 67 seasons (63 seasons for Florida State) whereby it appeared in the AP’s preseason poll (followed by, of those appearances, the number of seasons ranked and not ranked in final poll), and number of seasons appearing in the AP’s final poll (followed by, of those, number of seasons ranked and not ranked in preseason poll).

(For example, Georgia, although borderline moderate, has had a weak relationship at .395 since 1950 in regards to what the AP ranks the Bulldogs in the preseason as it relates to where they finish in the final poll. In the last 67 years, Georgia has been preseason ranked on 35 occasions: 22 times it finished in the final poll, 13 times it finished unranked. In the last 67 years, the Bulldogs finished ranked on 30 occasions: 22 times they had been preseason ranked, 8 times they had not.

I next figured the exact same as above, but for just the last 10 seasons (2007-2016) and, interestingly, found somewhat similar results with Clemson, Alabama, and Florida State all having strong relationships, Tennessee at nearly strong (.587), and Georgia (.313) and Auburn (minus-.263) again in the weak zone. This should be no surprise considering the Bulldogs were preseason ranked each of the last eight years (with an average ranking of approximately No. 13), yet finished ranked in the final poll just three times.

Finally, I again figured r for each of the nine teams beginning in 1950, but for only the seasons in which they were preseason ranked. Ranked according to r, each team is also listed with the number of 67 seasons (63 seasons for Florida State) whereby it appeared in the AP’s preseason poll followed by, of those appearances, the number of seasons it finished ranked as high or higher than its preseason ranking, and the number of seasons it finished ranked lower (including unranked) than its preseason ranking. And, notably, again the results are somewhat similar as before:

So, have preseason polls been a good indicator of where teams will rank in the end? When it comes to the AP Poll, and whether since 1950, just the last decade, or only when teams are preseason ranked, there is primarily a moderate-to-strong association between preseason and final rankings for most of the teams measured. In other words, the preseason AP Poll does kind of “matter,” and probably for most schools regardless of geographic location. However, for some teams, like apparently Georgia—and who knows why this is the case—preseason polls are like…