Pollsters Overreact to Teams Like Texas and Wisconsin With Big Week 1 Wins

Wisconsin managed to somehow contain LSU’s offense on Saturday, dominating the #5 team in the preseason poll at Lambeau Field 16-14 on Saturday. Texas nixed Notre Dame in impressive fashion, beating the #10 ranked Notre Dame Fighting Irish 50-47 in OT at home.

For that, both teams made huge, almost unprecedented jumps in the AP Poll. Both were unranked in the initial polls but now Wisconsin is at #10 and Texas is at #11. (We’ll get into just how almost unprecedented in a moment).

Those moves were just part of a first week with plenty of games that challenged our conventional wisdom. On an opening weekend that had plenty of big conference matchups, several teams moved up and down after the dust settled. Houston, Georgia, Washington, and Louisville made sizable jumps in the polls. Oklahoma, Tennessee (after surviving in OT), Notre Dame, Ole Miss, and LSU fell down at least 8 spots each. UCLA, North Carolina, and USC fell out entirely.

This tweet got me thinking:

How much more do we know after one week of games? This much is certain. We think we know a lot more. But is that actually true?

I went back through past AP preseason polls, the polls after the first game for each team, and the final AP poll to compare.

Let’s start with teams that are unranked and make big leaps into the polls after the first game. Using the historic poll information at sports-reference.com/CFB, here is every occasion where an unranked team made at least a 10 spot leap in the poll after the first game. A couple of notes here. Before 1968, the AP only ranked the top 10, which is why I stopped there. Starting in 1989, the AP began the top 25, and before that, it was a Top 20. For the years 1968-1988 then, it only includes previously unranked teams that entered the poll at 11 or higher.

Just so you are reading that correctly, Texas A&M was unranked last year, beat Arizona State in week 1, and was ranked #16 the next week. They finished the year unranked.

The last time a team shot up the polls as much as either Texas or Wisconsin did this week after one game, after being unranked, was back in 1976 and 1975, when Missouri upset USC and Alabama on the road in consecutive years.

Four of the seven unranked teams who shot up the polls finished the year unranked. None of them finished closer to their week 1 poll ranking, or exceeded it.

Let’s expand it out, to not just unranked teams, but other teams that shoot up. Here’s a list of every other team that was ranked in the preseason, but moved up 10 or more spots after the first game. Same caveat applies, with only a Top 20 being used before 1989.

Of those teams, three finished with an even better ranking than where they rose after week 1: Alabama in 2008, Illinois in 1989, and UCLA in 1985.

If we combine the two lists, we find that 14 of the 21 teams (67%) finished the year worse or closer to their preseason ranking (when ranked). More often than not, the week 1 result led voters astray of how good the teams were.

Okay, so set apart those extreme jumps. How much more do we know after week 1 than in the preseason? I looked at the last five years of AP preseason, week 1, and final polls. I compared the absolute value difference for all teams that appeared in at least one of them, between (a) the Final poll and Preseason poll, and (b) the Final poll and week 1 poll. (I used a rank of 30 for any team unranked in any of those polls).

The Week 1 AP Poll, after we’ve seen a slate of results, has been ever so slightly less accurate than the Preseason AP Poll, when comparing both to the final season results. The average team in the Preseason poll is 9.8 slots away from final ranking versus 9.9 slots for Week 1.

Okay, so let’s call it a wash. We think we know more, but we don’t. Sure, last year pollsters correctly identified Arizona State as unworthy and dropped them. They replaced them with Texas A&M, who beat them. They correctly moved Utah into the Top 25, but that was more than counteracted because Stanford, who would finish the year in the Top 5, was dropped from the poll after the loss at Northwestern.

There’s a good chance that when we look back at this season, the preseason poll was again a more accurate gauge of what would happen, compared to just looking at the first result in a sport where there is no preseason. Not that you shouldn’t be impressed that a team that was a 3.5 point underdog and playing at home managed to go to overtime, but we are often lacking in perspective with just one result, and conventional wisdom is there for a reason.