Neutral Field
Public college football power rankings where every team's rating breaks into parts that add up to the published number.
Every FBS team, ranked by how much it would beat an average FBS team on a neutral field. All 138 of them. The ranking comes from a frozen model inside Bet Advisor, and it doesn’t change its method mid-season, so a team’s rank never moves on a vibe.
The rating explains itself. Each team’s score is this season’s Elo plus last season’s opponent-adjusted efficiency edges, and the team page shows every part with its sign. Indiana’s +26.34 is Elo, success rate, EPA per play, and six smaller edges, some of them negative. Most published ratings don’t show you that.
The front page is “if the season ended today”: a 12-team bracket seeded from the ranking with conference champions taking the automatic bids, the last four in and first four out, and a toggle for the 24-team format reported for 2027. A projected view plays every remaining game 10,000 times with the win probability implied by the power gap. Around that sit quality tiers cut wherever consecutive scores differ by a point or more, quadrant résumés, strength of schedule, conference standings, a week-by-week bump chart, and a page of insights like poll disagreements and one-score luck.
There’s no backend. A Python exporter writes a JSON bundle with every rule spelled out in it, the Sunday grading run commits it, and Cloudflare Pages deploys the static site on the push. A re-export with the same inputs is byte-identical, so the weekly commit only carries what changed.
It went live October 28, 2026, at cfb-rank.brentgann.io. It’s for group chats arguing about the playoff on Saturday night, so it’s built for a phone first.
Every team page shows what makes the rating. The parts sum exactly to the published number.