We ranked all 134 FBS teams in May. Here's what that's worth.

It's May. Spring practice is closed, the transfer portal is mostly done its damage, and no FBS team has published its 2026 roster yet. Most won't until late July. Camp doesn't open for three months. The first real game is on August 29.

We ranked everyone anyway.

The 2026 Way Too Early Top 25 is live on the front page, and the full 134-team projection is at /rankings/all. The label matters: this is what GIE thinks an honest, evidence-based ranking looks like with the information available today, and it's labeled "Preliminary 2026" everywhere on the site for that reason. It's not the model speaking with its full voice. It's the model speaking with two of its three primary instruments.

What the number is made of

Three inputs go into the preliminary rating:

Prior-season finishing strength. Every team starts with where they ended 2025. This is by far the single most predictive variable in a preseason projection — a great team last year is overwhelmingly likely to start strong this year, and a bad team has a lot to prove before climbing. Football just isn't that volatile season-to-season at the team level.

New-head-coach drag. If a program hired a new HC over the offseason, GIE applies a small penalty in their first year. Historical data across the last decade says Year-1 HCs run roughly 0.19 wins below what their team's prior strength would predict. That doesn't sound like much, and it isn't, but in a sport where five wins separates the top of the SEC from the bottom, it's enough to shift teams meaningfully in a tight band.

2026 recruiting class. Top 5 classes get a modest bump. Classes outside the top 50 get a modest haircut. Recruiting matters, but most of its impact lands two or three years out as those classes develop. The Year-1 signal from a single class is real but small, and we calibrated the adjustment accordingly.

That's the whole equation today. Three inputs, transparent weights, every team gets the same treatment.

What's deliberately missing — and why it matters

The single biggest absence is returning production: how much of last year's offensive and defensive production is back on the roster this year. The QB room. The offensive line. Returning starters at skill, in the front seven, in the secondary. Public research has been clear for over a decade that returning production is the second-biggest preseason signal behind prior strength.

We don't have it yet. CFBD — our primary data source — doesn't publish finalized 2026 rosters until teams release them in July. Until then, "who's actually coming back" is a guess on every team in the country, and GIE doesn't make guesses it doesn't have to. So returning-production stays out of the preliminary number.

A second missing piece, smaller but real: coordinator continuity. When a team hires a new OC or DC, the impact can be material — sometimes more material than a new HC. We have the framework to model it, but our coordinator data for 2026 is still being assembled.

What's coming when rosters publish

In late July, when CFBD's 2026 roster data lands, GIE re-rates every team using its full toolkit. The Way Too Early ranking on the homepage gets replaced with the actual preseason projection — the one we'd stake the model on.

Expect movement. Some of it will be dramatic.

Teams that quietly returned a Year-2 starting quarterback and four offensive linemen will climb, sometimes ten spots or more. Teams that lost a starting QB to the portal in March and haven't found a replacement will fall hard. Programs whose new head coach also rebuilt the coordinator staff will face a compounding drag. Programs whose new head coach inherited continuity will get a different reading.

That August ranking is the one to take seriously. This one is calibrated for May. It's directionally useful for offseason arguments — is the preseason Top 10 about right?, who's overrated based on last year?, which new-HC programs are the system worried about? — and not for much else.

How to read it in the meantime

Look at the bands, not the lines between teams. A team ranked 8th and a team ranked 12th in May are essentially indistinguishable; the model can't separate them at this resolution with the data it has. Differences of 20+ spots are meaningful; differences of 3-4 spots aren't.

Look at the new-HC tags, too — those are the teams whose number carries the most baked-in uncertainty in the current build, and they're also the teams most likely to move when rosters confirm.

And come back in August. The number you'll see then is the one this site was actually built to produce.

— 4thand4cast. College football, by the numbers.