Forde-Yard Dash: There’s a Major Flaw With Early College Football Rankings

Forty names, games, teams and minutiae making news in college football, where early-season firings are back in vogue at UCLA and Virginia Tech.

First Quarter: But Remember, the Selection Committee Is the Problem

Most teams are 25% of the way into the regular season. It’s still early in the College Football Playoff race, but a not-insignificant amount of results have been acquired.

Have you seen the analytics ratings? Oh boy.

Alabama (1) is in the top five in almost all of them. Florida State (2), which thumped the Crimson Tide by two touchdowns in front of 10.7 million viewers—we all saw it—is outside the top 25 in almost all of them. ESPN’s Football Power Index, Bill Connelly’s SP+ ratings and the Sagarin Ratings have the Seminoles anywhere from 27 to 33.

ESPN’s FPI also has Michigan (3) at No. 10 and Oklahoma (4), which defeated the Wolverines by 11 points Sept. 6, at No. 13. And 0–2 Notre Dame (5) is at No. 14 one day after losing at home to 3–0 Texas A&M (6), which is No. 15.

Given how viciously people went after the AP voter last week for her (admittedly very bad) Top 25 ballot, shouldn’t there be some scrutiny of what the computers are spitting out? If we’re going to get judgy, let’s judge all the rankings.

There is an easy but unsatisfying explanation for what the analytics rankings are putting out: Most of them factor in past performances from previous years in building a preseason projection. Connelly at least that disclaimer and detailed explanations of how his rankings work in his weekly updates on ESPN.com. (His “resume” rankings, which differer from the SP+ predictive set and are lower on the page, have Florida State No. 2 and Alabama No. 38.)

The inherent statistical biases built into many preseason projections were these: Florida State, coming off a 2–10 disaster, assuredly had very low power rankings. Oklahoma was 6–7 last year, which diminished its projection. Notre Dame has certainly performed much better in recent years than Texas A&M and thus started out ahead of the Aggies.

Unlike human voters, who can watch the results on the field and make radical revisions to preseason rankings, it takes time for the old analytics data to rinse out as new data comes in. The question is, how much time before the power rankings are an accurate indicator of, you know, actual current power? If it takes longer than three weeks—which it clearly does—should ESPN and other outlets hold off on publishing incomplete information?

That won’t happen, because everyone loves rankings—they stoke excitement and debate. The same argument is often made against the human polls being released in the preseason, and the rationale is the same: Rankings are good for business.

Sports Illustrated also published a preseason Top 25 and playoff bracket, because we like reader traffic as much as the next media outlet. The challenge is to course-correct once actual games are played, and to leave August guesswork behind.

Which is why The Dash playoff bracket is yet to include our preseason No. 1 team, Penn State (7), which to date has beaten Nevada, Florida International and FCS program Villanova. (The 52–6 victory over ‘Nova was suitably impressive for SP+ to vault the Nittany Lions up six spots, to No. 1.) James Franklin’s team has to play somebody of substance before it earns a Dash bracket spot.

While the Nittany Lions remain out for at least two weeks—they’re idle Saturday before playing Oregon on Sept. 27—The Dash this week welcomes perennial non-power Vanderbilt (8). The Commodores have defeated a pair of Power 4 conference opponents—Virginia Tech and South Carolina—on the road by a combined 48 points. They have been a dominant team against peer-level competition.

Vanderbilt celebrating a win over South Carolina. — Vandy is off to a great start with a couple of high-profile wins. | Jeff Blake-Imagn Images

That’s enough to host a first-round playoff game, should the playoff start today. Modern Vandy shouldn’t be penalized by historical Vandy.

The important concept for human voters to keep in mind: no slot voting. Starting the year with Team X ahead of Team Y and keeping it that way just because they both keep winning fails to account for who they’re beating and how impressively. Starting with as blank a slate as possible each week helps.

Ultimately, if the analytics all catch up to the work done this season, none of the September rankings matter very much. But along to the way to figuring it out, there is a risk of outraging fans and undermining credibility. If ESPN is going to publish an FPI that makes no sense—exacerbating pre-existing, pro-Alabama conspiracy theories, especially in Tallahassee—it has to deal with the blowback.

And this issue has to be viewed within the context of what the power cabal running the sport is pushing for. Namely, less “subjectivity” from the CFP selection committee, and a greater reliance on formulas. Commissioners Tony Petitti (9) of the Big Ten and, to a lesser extent, Greg Sankey (10) of the Southeastern Conference, have put pressure on the committee to adjust its modus operandi to fit their liking.

Petitti has gone so far as to push the ruinous idea of multiple automatic bids per conference—ostensibly as a method to reduce committee decision-making. Sankey is less inclined to go that route, but has hammered home “strength of schedule” as a deciding factor. The committee’s selection criteria has accordingly been adjusted

Both men have been careful not to outright assail the committee members, but they’ve passive-aggressively stepped up to the edge of it. Petitti especially.

The basic list: We have great confidence in the selection committee, as long as it does what we want them to do in regard to appraising our conferences. If they don’t, we’ll take it out of their hands as much as possible in 2026.

What would be taken out of the hands of a human committee would, in theory, be dropped into the hard drive of computers. That wouldn’t make the results infallible. Even in an AI world, there is a need for human insight, interpretation and deliberation.

Unless, of course, everyone is cool with Alabama being ranked 25 spots ahead of an undefeated Florida State team that whipped the Crimson Tide two weeks ago.

Updated Dash College Football Playoff Bracket Seeding:

Miami
Ohio State
LSU
Florida State
Oklahoma
Oregon
Vanderbilt
Texas A&M
Georgia
Georgia Tech
Utah
Tulane

First-round games:

Tulane at Oklahoma
Utah at Oregon
Georgia Tech at Vanderbilt
Georgia at Texas A&M

Quarterfinals:

Tulane-Oklahoma winner vs. Florida State in the Cotton Bowl
Utah-Oregon winner vs. LSU in the Sugar Bowl
Georgia Tech-Vanderbilt winner vs. Ohio State in the Rose Bowl
Georgia-Texas A&M winner vs. Miami in the Sugar Bowl

Also considered: Auburn, Missisisppi, TCU, Iowa State, Tennessee, Mississippi State, Missouri, USC, Notre Dame, Texas, South Florida, North Carolina State, California.