Rankrly Home
Rankrly Research
12-season matched studyPoll Research15 min read

Does Starting Ranked Give You an Advantage?

A 12-season matched analysis finds that teams beginning in the AP Top 25 stay meaningfully higher in the poll than otherwise similar teams, but most of the gap disappears by November.

By Rankrly Research · Published September 2, 2026

A preseason ranking is supposed to be temporary. Once the games begin, the poll should have enough new evidence to judge teams on what they have actually done.

Here is the basic test. Take a team that was ranked before the season and pair it with a team that was not. The two teams must be from the same year, have the same FBS record, and rank almost identically based only on results. Then compare where AP voters placed them.

Three terms to know

Matched pair

Two teams from the same week with the same FBS record and nearly the same results-based ranking.

Ranking boost

How many AP poll places higher the preseason-ranked team sat than its matched opponent.

Top 25 gap

The difference between the two Top 25 appearance rates. Because both rates are percentages, the gap is measured in percentage points.

The answer

Yes, especially in September.

Across 2,002 matched pairs, the team that started ranked sat an average of 6.70 places higher in the AP poll. It appeared in the Top 25 63.24% of the time, compared with 22.63% for the similar team that started unranked. That is a 40.61-point gap.

This does not prove AP voters intentionally favored teams that started ranked. It shows the gap remained even after comparing teams with the same record and almost the same results-based ranking.

6.70

Average ranking boost

40.61-point gap

Top 25 appearance gap

2,002

matched pairs

12

seasons analyzed

Week by week

The advantage is huge early and nearly gone by Week 12

After Week 1, the team that started ranked sat 12.49 places higher than its matched opponent. After Week 12, the average difference was only 0.99 place.

How the preseason head start changes each week

A higher number means the team that started ranked received more of an advantage over its otherwise similar opponent.

After Week 5

6.96 places higher

Among 144 matched pairs, the team that started ranked sat this much higher in the AP poll on average.

Week 1Week 15

All 2,002 results

The adjusted gap favored the preseason-ranked team in 76% of pairs

Each bar shows how many pairs fell within a range of adjusted AP-place differences. Positive values favor the preseason-ranked team; negative values favor the preseason-unranked team.

76.2%

favored the preseason-ranked team

22.2%

favored the preseason-unranked team

4.00

median ranking boost

1.00 to 12.00

middle half of results

Values to the right of zero favor the preseason-ranked team. Values to the left favor the preseason-unranked team. Each bar shows the number of matched pairs in that range.

Starting position matters

Starting in the Top 10 brings the biggest head start

Preseason Top 10

11.38 places

677 matched pairs. The team that started ranked made the Top 25 in 81.83% of comparisons, versus 31.91% for the similar team that started unranked.

Difference in Top 25 appearances

49.93-point gap

Preseason #11–25

4.31 places

1,325 matched pairs. The team that started ranked made the Top 25 in 53.74% of comparisons, versus 17.89% for the similar team that started unranked.

Difference in Top 25 appearances

35.85-point gap

Season check

Teams that started ranked finished higher in every season studied

The size varies, but no single year creates the overall result. The unusual 2020 season has the largest estimate and remains in the main sample rather than being quietly removed.

2014

7.08

AP places

2015

5.36

AP places

2016

5.66

AP places

2017

6.94

AP places

2018

6.99

AP places

2019

7.67

AP places

2020

8.48

AP places

2021

5.41

AP places

2022

5.19

AP places

2023

7.06

AP places

2024

7.49

AP places

2025

7.20

AP places

Robustness checks

The conclusion survives several reasonable changes

These checks change one assumption or isolate one part of the sample. They show how sensitive the result is without presenting a new matching method as though it were the original analysis.

Published estimate

All matched pairs, with every team outside the Top 25 treated as position 26.

Matched pairs

2,002

Average boost

+6.70

Within one resume place

Tightens the published two-place matching limit to one place.

Matched pairs

1,581

Average boost

+7.27

Exclude the 2020 season

Removes the pandemic season while keeping the original matches from every other year.

Matched pairs

1,819

Average boost

+6.52

Weeks 1 through 5

Restricts the result to the opening part of the season.

Matched pairs

1,024

Average boost

+9.92

Weeks 10 through 15

Restricts the result to the later part of the regular season.

Matched pairs

491

Average boost

+1.93

Treat unranked as position 30

Keeps every match but moves teams outside the Top 25 four places lower.

Matched pairs

2,002

Average boost

+8.32

Treat unranked as position 40

Keeps every match but moves teams outside the Top 25 fourteen places lower.

Matched pairs

2,002

Average boost

+12.38

The position-30 and position-40 checks do not claim that every unranked team was actually 30th or 40th. They test whether the main conclusion depends on assigning all unranked teams position 26.

Matched-pair explorer

Inspect the comparisons behind the headline

Filter every matched pair by team, season, week, or starting tier. “Preseason-ranked” and “preseason-unranked” describe where each team began the season. The AP column shows its status in the poll for the week being compared.

Raw AP gap compares the two current poll positions. Adjusted gap accounts for the small difference between their results-only resume ranks. “Unranked” means outside the current Top 25.

Loading matched pairs...

Program lookup

See how one program fared in its matched comparisons

Pick a program to turn the analysis around and view every result from that team's perspective. Positive numbers mean the selected program sat higher than its matched opponent. Negative numbers mean it sat lower.

Choose any FBS program to see every time it appears in a matched pair.

Methodology

Match the record, then match the resume

The goal is to compare teams that look as similar as possible based on current-season results. The only major difference left is whether one team appeared in the preseason AP Top 25.

  1. 1

    Use every regular-season AP poll from 2014 through 2025 after the preseason poll, along with completed games between two FBS teams through that point in the season.

  2. 2

    Build a results-only Colley ranking for each season and week. It uses wins, losses, and opponent connections without preseason expectations, margin of victory, recruiting rankings, or conference weights.

  3. 3

    For each team that started in the AP Top 25, find one team that started unranked in the same season and week. The two teams must have identical FBS records and results-only Colley ranks within two places.

  4. 4

    Use each control team once per season and week so one unranked team cannot support several comparisons.

  5. 5

    Treat a team outside the AP Top 25 as position 26, then subtract the small remaining resume-rank difference from the raw AP-position gap.

  6. 6

    Average those matched differences overall, by week, by preseason tier, and by season.

What one match means

Same season. Same poll week. Same FBS wins and losses. Results-based rankings no more than two places apart. Each unranked comparison team is used once.

Limits

What this result does not prove

The analysis finds a pattern, not cause and effect. The result looks like poll inertia, but it cannot prove the preseason ranking caused the later gap.
The Colley model intentionally ignores margin of victory and other context voters may reasonably use. Matching teams within two resume places narrows the comparison without making the resumes identical in every respect.
Encoding every unranked team as position 26 compresses differences below the poll cutoff. The analysis measures Top 25 position, not how far an unranked team sits from receiving a vote.
The 2020 pandemic season had an unusual schedule and still remains in the main sample for transparency. The estimated advantage is positive in every individual season, not only in 2020.
The study follows regular-season AP polls and does not claim the same effect size for the final postseason poll or the College Football Playoff selection process.

Sources and reproducibility

Polls, games, and a resume-only baseline

The published dataset contains 22,452 observations. One observation is one team in one AP poll week. Those observations produced 2,002 matched pairs. FCS games are excluded from both the records and the results-based ranking.

Bottom line

Starting ranked looks like a real head start, not a season-long guarantee.

The preseason poll matters most when voters have only a few games to judge. By November, actual results have erased most of the difference. Starting ranked helps a team get noticed early, but it does not protect that team all season.

More Rankrly Research