🏈 No, preseason college football polls don’t impact the Playoff like you might think - NY Times

College football’s preseason polls are under attack. The Big 12 ended its media poll this year, apparently assuming formalized expectations could do more harm than good. Conference USA didn’t do one, either. The Big Ten hasn’t had one for years, and there were conversations at its media days about whether preseason rankings carry more meaning than they deserve, influencing the opinions of a College Football Playoff selection committee that doesn’t start meeting until later in the season.

Do they have a point? Do muted forecasts from preseason polls artificially lower a team’s ceiling in the College Football Playoff race? Is a team like 2024 Big 12 champion Arizona State weighed down if it was picked to finish last? Are the teams snubbed in the Associated Press preseason poll, which will be released Monday afternoon, at a disadvantage by the Playoff selection committee or improperly buoyed if they did make the list?

If so, we can’t find any evidence to back it up.

We looked at five seasons from the College Football Playoff era, scrutinizing preseason rankings against final decisions. We also averaged four sets of advanced metrics: the Massey ratings, Sports Reference’s Simple Rating System (SRS) and ESPN’s Football Power Index (FPI) and SP+. The computer metrics are intended to be predictive, while pollsters tend to focus on teams’ results and resumes. But because the computers use formulas rather than people to rank teams, they’re a helpful (even if imperfect) counterpoint to human polls, whether they’re done by AP voters, coaches or the CFP selection committee.

If anything, our review shows the opposite of what critics claim: Teams with higher preseason rankings seem more likely to be punished, not rewarded, compared to the computer rankings, while teams that start unranked are more likely to be overrated than underrated.

2024​

It’s hard to claim Arizona State and Boise State were dinged for starting unranked. Both earned top-four CFP seeds and byes as conference champions ahead of Clemson (which began in the top 15 and was picked to finish second in the ACC).

If bias affected the committee, then, it would have come with the final at-large selections. Those went to Indiana and SMU. Neither was ranked in the preseason Top 25, and the Mustangs were picked to finish seventh in their inaugural ACC season.

The first two teams out in the CFP rankings were Alabama and Miami. End-of-year advanced metrics put the Crimson Tide and Ole Miss in the top eight, making them the computers’ biggest CFP snubs. All three of those teams started in the preseason top 20, with Alabama and the Rebels both in the top six. The early outside expectations, clearly, did not help those three or hurt the Hoosiers or Mustangs.

Study the final AP poll, and you’ll reach a similar conclusion. Thirteen teams started unranked but ended the year in the Top 25. If preseason rankings still carried over, we’d expect those teams to be devalued compared to their true quality.

Nope.

Of those 13, only two were ranked lower by the pollsters than by computers: No. 10 Indiana (one spot lower) and No. 19 South Carolina (eight spots lower). The other 11 were all overvalued by at least two spots. The Big 12’s representatives — Arizona State, Iowa State, BYU and Colorado — were a combined 34 spots higher in the poll than in advanced metrics. The Sun Devils were seventh in the AP poll but 24th in the computer rankings. On the other end, the team most undervalued by pollsters compared to computers was Alabama (10 spots), which began the season ranked No. 5.

Seven teams were in the top 25 of the computer averages but snubbed in the final AP rankings. Five of them, however, were ranked in the preseason poll but fell out. If there was a bump, LSU, Texas A&M, USC, Iowa and Michigan somehow missed it.

2023​

The most controversial decision of the four-team CFP era was picking 12-1 Alabama over 13-0 Florida State. Considering the Seminoles were four spots higher in the penultimate CFP standings, it’s hard to connect their preseason rankings (No. 4 Alabama vs. No. 8 Florida State) to that close call. Preseason No. 1 and SEC favorite Georgia missed the field, too, despite finishing second in advanced metrics.

In a 12-team bracket, the CFP’s final at-large bids would have gone to Ole Miss and Oklahoma ahead of LSU. The Sooners began the fall 15 spots behind the Tigers; the Rebels were two spots worse than that.

Again, teams that were ranked higher in the preseason were more likely to be undervalued by the pollsters. The biggest negative gap between advanced metrics and the poll was by the preseason No. 3 team, Ohio State (finished third in the computers and 10th in the AP poll). Of the teams in the computers’ top 25 but unranked by the pollsters, more began the season ranked (three) than unranked (two).

The biggest beneficiary from voters was Oklahoma State. The Cowboys were 16th in the final AP poll — 19 spots higher than the computers’ average. They apparently were not harmed by being unranked in the preseason or being projected to finish seventh in the league.

2022​

Being picked to finish seventh in the 10-team Big 12 didn’t preclude TCU from making the bracket as the No. 3 seed. Alabama was the preseason No. 1 team ahead of Ohio State, but the Buckeyes got the fourth and final spot ahead of the Crimson Tide.

A 12-team field with the CFP’s rankings would have featured five teams that were unranked in August: TCU, Tennessee, Kansas State, Penn State and Washington. The most overvalued Playoff team would have been USC (18th in the final computer rankings). Would the 11-2 Trojans have made the field because of an edge they earned by starting 14th in the AP poll? Or because they had two fewer regular-season losses than the first team out (Florida State) and beat the next team out (Oregon State) on the road?

The Big 12 does have the biggest gripe in our analysis. Texas began this season unranked and fielded a top-10 team according to advanced metrics, yet it finished only 25th in the AP poll. But it’s hard to feel too bad about the disparity considering Texas’ 8-5 record, even if three losses were to very good teams (Alabama, Washington and CFP finalist TCU).

2021​

Although Cincinnati started the season one spot ahead of Notre Dame, that’s not the easiest explanation why the Bearcats earned the fourth and final CFP bid over the Fighting Irish. A 13-0 record and win at Notre Dame suffices.

Using the final CFP rankings, half of a hypothetical 12-team field would have started the season outside the Top 25. One of them, ACC champion Pitt, would have taken the final spot. Three of the first four teams out were, unlike the Panthers, ranked to start the year: Big 12 preseason favorite Oklahoma, Pac-12 preseason favorite Oregon and No. 18 Iowa.

The strongest counterargument is from BYU, which went from unranked in the preseason to 13th in the final CFP rankings. The Cougars were 10-2 in the regular season (one win more than Utah) with a head-to-head win over the Utes, yet they still finished behind their rival. But the facts, again, don’t point to a preseason bias. BYU was ahead of Utah in the penultimate rankings, and the Utes jumped the Cougars only after blowing out No. 10 Oregon by four touchdowns in the Pac-12 championship game. Utah also finished the year ranked 13th in computer averages; BYU was 36th.

2014​

We’re skipping back to 2014 to revisit the only other true debate of the four-team era. Ohio State rose from fifth to fourth in the final CFP rankings after routing Wisconsin 59-0 in the Big Ten title game, leapfrogging co-Big 12 champion TCU (which beat Iowa State by 52 that week) and staying ahead of the other co-Big 12 champion, Baylor, which beat Kansas State by 11.

The Buckeyes started fifth in the preseason poll, five spots ahead of the Bears. TCU was unranked. Good luck tying those August rankings to the December decision.

When the CFP released its first rankings in late October, all three teams were 6-1. Last in the bunch: Ohio State. Why would preseason expectations have suddenly become a deciding factor a month and a half later? The outcome also validated the choice. Ohio State won the national title and finished higher than both Big 12 teams in the computer averages.

Final thoughts​

After spending seven full seasons as an AP voter, I can tell you that preseason polls do matter … for a few weeks. The disparate early schedules make it hard to weigh a team that kicked off against a marquee opponent vs. one that started against an FCS foe. Openers are weird, too, so it’s risky to make too much of them. But after the first month or so, August expectations wash out in favor of what actually happened, not what you thought would happen.

The numbers bear that out. We found no compelling case that preseason rankings made a material difference in any actual or hypothetical CFP decision, or that they made an outsized impact on the rankings.

Over the past four years, 50 teams finished ranked in the AP poll after starting the season unranked. Of those 50 teams, 36 were put higher by voters than by the computers; only 10 were underrated by the pollsters. That’s the opposite ratio of what would happen if teams were punished for low outside expectations.

Another set of numbers: 25 teams that would have been ranked in a computer poll finished unranked by voters. If preseason polls resonated through the year, we’d expect more of those undervalued teams to have been unranked in Week 0. That’s not what happened. More started ranked (15) than unranked (10).

Is it possible that summer polls and projections have smaller, less obvious impacts on the field or in the standings? Sure. But it’s more plausible that rankings are a just-for-fun way to get fans excited for the new season by giving them a chance to do what they do best: argue about how their rival is overrated (again).