Ratings Explained

The number

A team's Rating is the margin it would be expected to win by against an average NFL team on a neutral field. A +6 team should beat a +2 team by about 4 on a neutral field, or by about 5.5 at home, since home field is worth 1.5 points in the model. Zero is average. The best teams in a typical season land near +8 to +10; the worst near −8. The NFL is a tight league: the gap from first to last is about what separates the top team from the middle in college football.

Opponent adjustment

Raw season averages, such as points per game or EPA per play, say nothing about who they came against. Six games against the AFC West is a different season from six against the NFC South.

The Rating fixes that by treating every game as an equation. For each team in each game, the points it scored equal its offensive strength, minus the opposing defense's strength, plus half the home edge. Every game of the season goes into one system of roughly 550 such equations, solved together, so scoring 27 on the best defense in the league counts for far more than 27 on the worst. The same solve runs on EPA per play, which is what the style profile's Adj. EPA per play shows.

Adj Off is the points per game a team would score against an average NFL defense on a neutral field, and Adj Def is the points it would allow against an average offense. Adj Off minus Adj Def is the Rating. Lower is better on defense.

Points, not just EPA

Both systems are tested against real games every year, and the mix between them is chosen by how well each predicts the next week. In the NFL, points scored and allowed predict future scores at least as well as per-play efficiency once the opponent adjustment and the prior are in place, so the headline Rating currently rests on points alone; the EPA solve is shown on every team page as Adj. EPA per play, where it tells you how a team got there.

Preseason prior

Before a season starts, every team gets a prior rating: its final rating from the previous season, kept at about 57% of its value, since NFL teams regress hard toward average from one year to the next. That weight is fit on every season since 2011, which is also where the chain of ratings begins, so even the first published season starts from a real prior rather than a blank slate.

The prior is worth about 8 games of evidence, more than in college football because 17 games against a tight league leave more noise in a record. After week one it dominates; by midseason it is less than half the rating, and by the playoffs it is a small share. Games also fade a little with age (about 3% per week), so a team that changed in October is not anchored to September.

How good is it

The model is checked by walking forward through past seasons a week at a time: fit the ratings on every game before that week, predict that week's games, then compare with what happened and with the closing point spread. The closing line is the hardest benchmark in the sport, and the ratings are not expected to beat it, only to stay close.

Beating the spread about half the time is the expected result for an honest model; anything much higher would mean the test was leaking future information. Early-season weeks miss by a little more than the rest, which is the price of leaning on the prior.

This season so far

The same test runs on 2026 after every data refresh: each week's games are projected from ratings fit only on the games before it, then scored once they are played. Through Wk 8, 120 games.

Projections and win probability

An unplayed game's projected spread is the difference between the two ratings plus 1.5 points for the home team (nothing at a neutral site). The projected score splits that spread around what each offense would do against the other defense. The win probability comes from the spread and the model's own miss history: actual margins land within about 13 points of the projection, so a 3-point favorite wins about 59% of the time and a 10-point favorite about 78%. Once a game is played, the projection shown is the one published the morning of the game.

Every game carries a thrill score from 0 to 100: half the quality of the two teams, a third how close the projection is, and the rest how fast they play. The games page ranks each day by it.

Playoff odds

The Playoffs page simulates the rest of the regular season 10,000 times. In each run, every unplayed game is decided by the model's win probability, and every played game stands. The runs then go through the NFL's seeding rules: four division winners by record, the next three by record as wild cards, the top seed resting in the first round. A bracket is played out the same way (higher seed at home, the Super Bowl on a neutral field) for the conference and title odds. Where a playoff game has already been played, its real result is used.

Ties in the standings use the league's procedure in order: head-to-head, division record (within a division), conference record, common games (at least four for wild cards), strength of victory, then strength of schedule. The remaining official steps, which rank teams by points scored and allowed, are replaced by point differential, and a coin flip settles anything still level. Ties are not simulated in individual games (about one game in 250 ends tied) but real ties count half a win. A team shown as clinched or eliminated is one that made or missed the playoffs in every single run, which can trail the league's official clinching announcements by a game.

The division standings on each division page come from the same procedure applied to the games played so far, so the order you see is the one the league would use today, not a win-percentage sort.

What it does not know

Injuries (including the quarterback), suspensions, weather, rest, travel and motivation are not modeled. Garbage time is not removed. Home field is one number for every stadium and every season, including 2020, when it was worth close to nothing. Treat the ratings as a well-calibrated starting point, not a betting sheet.

Lines and differences

The Lines page puts the market spread and over/under beside the projected spread and total for every game of the week, biggest disagreement first. The numbers are captured once a day with the site's refresh, so they are that morning's lines rather than the kickoff lines, and a played game keeps its closing line, so a finished week reads as a record of what happened against the number.

What is the diff?

The projected spread minus the market spread. If the market has a team favored by 3.5 and the ratings make it 6.2, that is a 2.7-point difference toward the favorite. Totals work the same way: the two projected scores added together against the posted over/under, as Over or Under. Differences of three points or more are in bold.

Which side does the model like?

The team named in the Diff column, which is the side of the market line the projection falls on, and Over or Under for the total. The win probability is the model's chance for the favorite, from the projected margin and the 13-point spread of misses. Under it is the market's chance for the same team, from the moneyline with the vig removed (each side's implied probability divided by their sum), and the gap in percentage points. On a played game the projection shown is the one published the morning of the game, and a check or a cross records whether that side covered.

How often does the model beat the closing line?

About half the time: 52.6% this season (114 games) and 49.5% over 2016–2025, against the 52.4% it takes to break even at standard −110 odds. That is a coin flip, and it should be: the ratings measure how good teams are, and the closing line is the sharpest public number there is. Read the differences as places where the ratings and the market disagree, not as bets.

When do lines appear?

The market posts the coming week's games on Sunday night and Monday, so the week ahead is normally complete by the Tuesday refresh and the week after shows no line yet. Every game that has been played shows its closing line, back to 2016.

For information only, not betting advice: the site takes no responsibility for any wager placed on it.