Opening the edition…
The first weekly ranking appears once every FBS and FCS team in this season’s field has played two counted games.
The Rutkowski Index is a completely independent college football rankings project. Every Sunday, the exact same formula grades every FBS and FCS team in that season’s field. Swap the logos on two helmets and their ratings stay exactly the same. Poll votes, conference reputation, money, and our own opinions have no say in a team’s score.
In 1960, renowned physicist and chess master Arpad Elo developed a rating system to answer a simple question: If two players sat down for a game of chess, how surprising would the outcome be? Ten years later, the International Chess Federation (FIDE) adopted his system, and it has become the foundation of competitive chess. Today, we apply that same principle to football, asking the same question whenever two teams take the field. FIDE’s history of Elo
It’s the gap between two Power Elo ratings that gives the numbers meaning. At a neutral site, teams rated 1600 and 1600 start even. Against a team rated 1500, that same 1600 team would be expected to win about 64 times in 100. A win by the underdog gives the model more reason to revise both ratings.
Power, or Power Elo, is pure football. It measures how strongly a team has played. We compare a team’s final margin of victory with the margin its rating and its opponent’s rating would suggest. Against the same opponent, a 21-point win earns more credit than a 3-point win. A heavy favorite can win by three and still lose Elo. Each extra point matters less as a blowout grows. We give the home team a 20-point Elo advantage when setting expectations for that game. At a neutral site that adjustment is zero.
Each week the model works through this season’s results together, updating both sides of every game in small steps. Suppose your team won a close road game in September. By November, the opponent has beaten three strong teams. That gives us a better measure of the team you beat, and your September win can lift your rating even during a bye.
College football isn't just about how good a team is. It's about the games they've actually won and lost. A dominant team can still have a disappointing résumé, while a less talented team can build an impressive one by winning difficult games. R Score (résumé) measures how many wins a team earned above what a 1500-rated team would expect against the same schedule. Suppose that a team would win a particular road game one time in four. Winning that game adds 0.75 to the résumé. Losing it subtracts 0.25. We add those differences across the schedule, using each opponent’s current strength. That’s how two 9–1 teams can have different résumés.
We build the Top 25 by putting a team’s Power Elo and R Score on the same scale and giving each half the weight. The standings call that hybrid score RElo (Rutkowski Elo). The 25 highest scores make the front page. All Teams shows the rest of the same table. We sort on the full score before rounding it for display. School names determine display order only when scores tie exactly.
Compare RElo with AP, CFP, FPI, SP+ and Massey in our study of ranking disagreements and Power Elo’s pregame picks.
A team page shows Power Elo and R Score alongside their season ranks. The Power Elo in each game row is what the model knew at the end of that game day. Δ Power Elo is the change since the previous row, including movement from opponents’ results. The opener is measured against the team’s starting rating. The Impact column shows how much higher or lower the team’s Power Elo is at the season cutoff with that game included. We replay the whole league once per omitted game. These separate effects can overlap.
Weekly standings appear once every team in the season’s field has played two counted games. Every earlier result contributes to those standings. Arrows compare a team with its rank for the same metric on the previous Sunday when both editions are visible. A team entering the first 25 is marked New. Open a team to see which results changed its résumé and which schools it passed. Editions recovered from actual releases are labeled separately from historical rankings reconstructed from scores. Browse the Sunday editions
Every team opened 2003 at 1500. Each new season begins with the previous season’s closing power ratings, rescaled around 1500 with a standard deviation of 150 points. Returning teams keep their relative starting strengths. A new team starts at 1500, and everyone begins with a fresh résumé. From opening day on, the model revisits only that season’s games.
Our records count completed games between two FBS or FCS teams, including the postseason. “Results through” gives the Denver kickoff date of the latest game included in the edition. We filter the schedule and calculate the ratings from the published scores. Game data from SportsDataverse · CC BY 4.0