Sources & Credits

This is an independent fan project. It is not affiliated with the College Football Playoff, the NCAA, any conference or any school. Team colors identify teams; no logos or other marks are used.

Data

CollegeFootballData.com (CFBD): every game, score, site, venue, team division and conference, the AP poll, the committee (CFP) rankings, team colors, box scores and play-by-play. Thank you to its maintainers. Terms of use.
College Football Playoff official rankings: used to check committee rankings by hand and to correct CFBD errors (see Methods).
CBS Sports: the Nov. 26, 2019 committee ranking, which CFBD is missing.

See Data coverage for which seasons and data types are covered, updated with every data refresh.

This site publishes only tables we compute (records as of a date, ranks as of a snapshot, rest days, scheduling measures). It does not re-host CFBD's raw data. To work with the raw data, get a free key from CFBD and use its API directly.

Prior work we credit and build on

John A. Trono, Improved Linear Model (Saint Michael's College): a linear model trained to match the committee's selections. Reproduced from his 2020 and 2025 tech reports (SMC-2020-CS-001, SMC-2025-CS-001) with his published weights, as a baseline on the Committee Formula page.
Best Ball Stats, Analyzing the CFP Rankings (2020): a two-stage regression of committee rankings.
CFB Numbers: weekly predictions of the committee's rankings.
FiveThirtyEight, College Football Playoff model (retired): modeled the committee's behavior from past rankings.
Benjamin Lucas, An analysis of the NCAA college football playoff team selections using an Elo ratings model (arXiv, 2024).
JThom Analytics: season simulations from a composite of rating models, with playoff projections and team comparisons.
IowaHawkBlog's Marble Game and the NEMindex: open, rules-based rankings (Marble Game rules).
Juyong Park and M. E. J. Newman, A network-based ranking system for US college football (2005).
The NCAA basketball RPI and NET, whose separate weights for home and road wins inspired our "wins by site" measure.

Other rankings worth comparing

Independent sites that rank teams their own way. We link to them; we don't copy their data.

College Football Marbles: the Marble Game, where teams win "marbles" from the teams they beat.
JThom Analytics: projections and side-by-side team comparisons.

Software

DuckDB and DuckDB-WASM (MIT): queries run in your browser.
pandas and Apache Arrow: the data pipeline.

Privacy

No accounts, no cookies, no tracking. Your case files, settings and favorite team stay in your browser.

Case Files can publish a case and let fans vote on it. Here is everything that involves:

A case you publish: what you typed in it (title, signature, notes and cards) is stored on our server so others can read it. You can unpublish it from the same browser.
Edit links: the key that lets you update or unpublish your case stays in your browser. We keep only a scrambled copy to check it.
Votes and reports: the first time you vote or report a case, your browser makes a random device id. It is sent only with votes and reports. Our server stores it only as a salted hash, and uses it only to count one vote (and one report) per device. It is never logged or joined to anything else. A report also stores the reason you typed. Your browser also remembers which cases you voted for, so the button shows it; that list never leaves your browser.
To limit spam, the server counts publishes, votes and reports per connection per day, using a salted hash that changes daily. Addresses are not stored.