AP Preseason Poll vs the Numbers, 2026
The AP preseason top 25 came out August 17. We took every team in it and looked up where the same team lands when our formula, the EL Preseason Rating v1, ranks all 136 FBS teams by rating. Most of the poll survives the check. Five teams do not, and the widest miss is not close: BYU is No. T14 in the poll and No. 26 on our board, twelve spots apart, outside our top 25 entirely.
The other two double-digit disagreements run the same direction. Houston polls 23rd and rates 34th for us. SMU polls 19th and rates 29th. The formula is also higher than the voters on a few names: USC, Tennessee, Texas A&M, Utah all rank better on our board than in the poll, with USC the largest of those at No. 9 for us against a T14 in the poll.
A rank gap is a disagreement, not a pick. When a disagreement is big enough to price, it shows up on the futures board with the number and the case attached.
All 25 poll teams · poll rank vs formula rank
Poll: AP Top 25 preseason, published Aug 17, 2026 (BYU and USC tied at 14). Formula rank: dense rank by rating over all 136 rated FBS teams, EL Preseason Rating v1. Gap = formula rank minus poll rank; 9 teams sit 5 or more spots from their poll position and carry the flag.
| Poll | Team | Formula rank | Rating | Gap |
|---|---|---|---|---|
| 1 | Ohio State | 1 | 31.2 | 0 |
| 2 | Oregon | 3 | 26.2 | +1 |
| 3 | Georgia | 2 | 26.4 | -1 |
| 4 | Notre Dame | 5 | 24.2 | +1 |
| 5 | Texas | 7 | 19.8 | +2 |
| 6 | Indiana POLL HIGH | 11 | 17.0 | +5 |
| 7 | Miami | 6 | 20.8 | -1 |
| 8 | Texas A&M | 4 | 25.5 | -4 |
| 9 | Ole Miss | 8 | 18.5 | -1 |
| 10 | Oklahoma | 10 | 18.1 | 0 |
| 11 | LSU POLL HIGH | 16 | 14.1 | +5 |
| 12 | Texas Tech | 13 | 15.6 | +1 |
| 13 | Alabama | 12 | 16.5 | -1 |
| T14 | BYU POLL HIGH | 26 | 10.9 | +12 |
| T14 | USC POLL LOW | 9 | 18.4 | -5 |
| 16 | Michigan | 14 | 15.2 | -2 |
| 17 | Washington | 18 | 13.9 | +1 |
| 18 | Penn State | 21 | 12.7 | +3 |
| 19 | SMU POLL HIGH | 29 | 9.1 | +10 |
| 20 | Tennessee POLL LOW | 15 | 14.5 | -5 |
| 21 | Utah | 17 | 14.0 | -4 |
| 22 | Iowa POLL HIGH | 27 | 10.8 | +5 |
| 23 | Houston POLL HIGH | 34 | 5.4 | +11 |
| 24 | Louisville POLL HIGH | 30 | 7.1 | +6 |
| 25 | Missouri | 25 | 11.0 | 0 |
How the formula ranks
Read the full methodology
One rating, all 136 teams, the same treatment. The EL Preseason Rating v1 scores every FBS team from five components in our database: last season's performance, four years of blue chip recruiting, returning production, transfer class quality, and coaching continuity. The rank on this page is a dense rank by that rating across all 136 rated teams, computed in the database, so a poll team's formula rank can fall anywhere from 1 to 136. The full rating table, with the components and the validation behind it, is at power ratings.
The model was tested before it ranked anyone. It was fit on the 2021 through 2023 seasons and held to two seasons it never saw, where it missed real win totals by 1.88 wins per team on average. That error bar is why we read a two spot rank gap as nothing and a twelve spot gap as a real disagreement.
What a gap means here. The poll is a panel of sportswriters voting on reputation, schedule memory, and what they watched in August. The formula only sees the database. When the two disagree by double digits, one of them is going to look wrong by November, and the season grades both of us in public. The win total version of this table, our expected wins against the posted market line for every team, is at win total predictions.
Provenance: poll side hand captured from the published 2026 AP preseason top 25 (Yahoo Sports · Sports Illustrated, both Aug 17, 2026). Formula side computed live into the page's dataset by scripts/el_poll_gap.mjs: dense_rank() over rating on edgelabs.el_preseason_rating (season 2026, model_version v1, 136 teams), generated 2026-08-20. Related: power ratings · win total predictions · futures board. Research and context, not picks · 21+.