College Football Conference Polls: How Accurate Are They?

College football media days have begun, and with them, conferences’ official preseason polls.

The preseason polls aren’t entirely meaningless. In addition to being a good press release, they reflect conventional wisdom entering the season. They’re also decently accurate.

We went back today through the four pre-pandemic seasons and compared these official preseason polls to the eventual results. Not every conference predicted a champion every year, with Conference USA and the Mountain West opting to only predict division winners and the Big Ten not running an official poll in 2016 or ’17. Overall, then, thirty conference champions were predicted, seventeen of them accurately, for a success rate of 57%. With divisions, the rate wasn’t much worse: 31 for 64, a 48% clip.

The polls get stronger when you limit them to the Power Five. Of the eighteen predicted conference champions in those leagues between 2016 and 2019, twelve were correct. With division champions, though, the margin isn’t so high: 15 for 28, or 54%.

This isn’t an evaluation of how smart media and coaches are in these conferences. Instead, it’s a gauge of how predictable college football is. In the Big 12, the poll routinely (and correctly) picked Oklahoma each of the four seasons. In the ACC, the poll was right three of four times, opting wrongly for Florida State in 2017. The Big Ten went one for two, with an errant Michigan pick in 2019. The Pac-12 went two for four. The SEC also went two for four, despite Alabama’s appearance in three of the four playoffs in question. On the one hand, these numbers are high. Two thirds of the time, according to our very small sample, you can expect a Power Five preseason poll to be correct. But on the other, they remind us that even in a sport dominated by the same few teams so routinely, there’s at least a moderate amount of uncertainty, even in Nick Saban’s own kingdom.