October 22nd, 2026
Elo, the chess formula we applied to NFL team strength in late 2025, works on a simple idea. Every result moves a team’s rating up or down, and the size of the move depends on how surprising the result was. Beat a stronger team and the number jumps. Beat a weaker one and it barely moves. How well that works depends on how much evidence the formula gets. An NFL club plays 17 regular-season games, one a week from September into January, so every result carries a lot of weight and a single upset can reshape a team’s rating for a month.
Baseball sits at the other extreme. An MLB club plays 162 games between late March and late September, close to one a day, and every game gets a price, a starting pitcher, a total and a run line. That daily rhythm built an entire consumer category in the United States, including guides to the best MLB betting sites that compare which operators cover the sport and what markets they carry. Sports betting in the US runs on state law, with separate licensing rules and minimum ages in each state, and anyone who needs help can call 1-800-GAMBLER. Football never had that volume, and the way its ratings move reflects it.
Seventeen games leave little room for error
With so few games, a rating system cannot wait for the noise to settle. Our football version answers this by moving ratings more per game than a baseball model would, and by carrying part of each team’s rating into the next season instead of starting everyone from 1,500, the midpoint in the Elo scale.
The trade-off is plain. A fast-moving rating catches a real change in a team quickly. In 2008, when the Patriots started the season with one of the highest starting Elos ever, it was built largely on Tom Brady playing as the starting quarterback. But a Week 1 injury that forced Matt Cassel into the starting lineup significantly changed the performance of a team that a slower rating model would not have detected.
A hard cap promotes parity
The league’s structure pushes teams toward parity, and Elo ratings follow suit to the middle. In 2026 every club works under a hard salary cap of $301.2 million and carries 53 players on the active roster, plus 16 on the practice squad. A hard cap ensures that no team can simply buy a deeper roster than its rivals, and the draft hands the best young players to the teams that finished worst.
For a rating system, that forced parity keeps the gap between the top and bottom of the league narrower than in sports without a cap.
The draft keeps adding new inputs
Each spring the draft changes rosters faster than any rating can follow. Since 2017 the International Player Pathway has added another layer, bringing in players from outside the American college system. Two of these players were drafted on April 25, 2026. Miami took the British tight end Seydou Traore in the fifth round, and Philadelphia picked Uar Bernard, a defensive end from Nigeria, at No. 251 overall.
None of those offseason moves can show up in a rating until the player takes the field, which is one reason early-season numbers deserve some caution.
Neutral sites test the home-field adjustment
Our internal Elo model gives the home team a fixed bonus, and that assumption gets harder to apply as the league plays more games abroad. The first regular-season game outside North America came in 2007, when the New York Giants beat the Miami Dolphins 13-10 at Wembley Stadium in front of 81,176 people.
The 2026 international schedule set a record with nine games across four continents, seven countries and eight stadiums. It opened at the Melbourne Cricket Ground on September 10, where 100,021 people watched San Francisco beat the Los Angeles Rams 27-7. For these games, the home-field bonus in a rating model has to be removed, because neither team is really playing at home.