Fathomratings

The numbers behind the number
2026 Season · Week 6
Player Ratings 1.0
1247 rated players
Player Ratings Player Projections
Player Ratings — production, measured against the defences actually faced

What the number means

FPP is Fathom Player Points per game: what a player's production has been worth in points, after subtracting what the defences he faced give up to everyone else. FPS — the Fathom Point Scale — places that value on a fixed 0–10 scale against every rated player at his own position since 2022: 5.0 is the median starter, 8.0 the 90th percentile, 10.0 the edge of the record. It is the same scale the team ratings use, read within a position rather than across all of FBS.

Compare quarterbacks to quarterbacks. A completed pass is credited in full to the thrower and in full to the catcher, so these numbers are honest within a position and meaningless added across a roster.

#PlayerConfFPSFPP ProductionGP

Method

The team rating solves for the numbers that best explain every margin once schedule strength is removed. Run that same solver on yards and touchdowns instead of margin and it returns, for every defence, how much it gives up above or below par — adjusted for the offences it has faced. Subtract a defence's figure from a player's line, in proportion to his share of his team's production that day, and what is left is measured against a common opponent. Yards and touchdowns are then converted to points by weights fit from four seasons of team results, not borrowed from a fantasy table.

A player with three games has told us little, so every rating is anchored to his rate from last season, with the anchor fading as games accumulate. A player with no last season starts at replacement level rather than at average, marked NEW in the table — he has to play his way up rather than being handed a number he has not earned.

None of this was assumed. On 38,570 held-out player-weeks across 2023–2025, predicting each player's next game: the raw per-game average missed by 3.94 points, the opponent adjustment cut that to 3.85 (p < 10⁻²¹), and the anchor to 3.80 (p < 10⁻⁴²). Both held in every season and at every position taken separately. Replacement-level priors beat average priors for newcomers (3.37 vs 3.45, p < 10⁻²²).

No betting lines are used anywhere in this rating, and it is not a betting model — our own prop models were measured against real posted lines and lost to the market. This is a description of what has happened. Games played is shown for every player; a rating built on two games deserves less of your confidence than one built on ten.

Player Projections — week 7, projected opportunities

Why opportunities, and not yards

A stat line is two things multiplied together: how many chances a player gets, and what he does with each one. We measured both across four seasons. A player's own prior efficiency predicts his next game's efficiency worse than simply using the league average — R² of −0.08 for receiving and −0.10 for passing. His prior opportunity volume predicts the next game's volume at +0.23 to +0.32.

So this page projects the half that can be forecast and leaves the half that cannot. Projecting receiving yards would multiply a real signal by a random one and present the result as a forecast. Carries, targets-per-se and pass attempts are the honest quantities, and they are also the ones that decide a stat line.

Players projected

2,034

Across the 121 of 138 FBS teams playing in week 7. A player needs at least one completed game this season to appear.

Pass attempts

224

From a trained model. It beats the player's own average by 5.7% and does so in every usage quartile — the only one of the three that earns the model outright.

Shown as an average

2,159

Figures where the model was measurably no better than the player's own prior average, so the average is what is printed. Marked ○ in the table.

PlayerConfOpponent Pass Att Carries Rec Rec line GP
Pass Att — trained model, all usage levels Carries — model above 9 carries a game, prior average below Rec — prior average○ Rec line — the posted over/under where a book has one. Our figure is an average and the line sits near the midpoint, so the two are not directly comparable and we publish no difference between them. Books post no line at all for carries or pass attempts in college football.

Sportsbooks post these same three quantities under different names: Carries is listed as rushing attempts, Pass Att as passing attempts, Rec as receptions. Same number, different label — worth knowing if you are reading this beside a posted line.