Positive and negative regression, explained

Opportunity carries forward. Efficiency mostly doesn't. A player scoring far more than his targets and carries usually produce is due to cool off, and a player scoring far less is due to bounce back. That is regression, and it is the most reliable edge in fantasy football.

What regression means

Every target, carry and pass attempt has a typical value. A target at the 5-yard line is worth more than one at midfield, and a deep throw is worth more than a screen. Add up a player's chances and you get what an average player would have scored with them: his expected points. Actual points minus expected points is his efficiency.

• Negative regression: he is scoring well above his chances. Expect fewer points ahead, even if his role stays the same.
• Positive regression: he is scoring well below his chances. Expect more points ahead, even if nothing about his role changes.

Regression pulls a player back toward his own level, not toward average. The best players beat their expected points every year. The question is how far above their usual level they are running right now.

Signs a player is due to cool off

• Touchdowns well ahead of his targets or carries, especially on long plays from outside the red zone.
• Big yards per target or per carry on modest volume.
• A quarterback throwing touchdowns on an unusual share of his passes, or almost never throwing interceptions.
• Points coming from a few long plays rather than steady work.

Signs a player is due to bounce back

• Lots of targets or carries, including near the goal line, but few touchdowns.
• A low catch rate on a high share of his team's targets.
• A quarterback with strong yards per attempt and a run of interceptions.
• Steady weekly work with no points to show for it yet.

Which stats carry over

We compared each player's first nine weeks with the rest of the season in 2024 and 2025, for every player with a real workload. "Carries over" shows how closely the first half predicted the second.

The pattern is clear. Role stats carry over: target share, touches and quarterback rushing. Rate stats mostly don't: touchdowns per chance, yards per carry and interceptions swing from half to half. Quarterback pass attempts are the one volume stat that barely carries over, because they rise and fall with the score of the game.

What happened to the hottest and coldest players

This season we split the first weeks from the next ones and followed the players scoring furthest above and below their expected points.

The hot running backs and receivers gave back nearly all of it. The cold ones climbed back to about even. Quarterbacks kept a little, because rushing and top-end passing are partly real skills.

How to use it

• Sell the touchdowns, buy the targets. A receiver with two long scores on four catches is a sell. A receiver with ten targets and no score is a buy.
• Trust quarterback rushing. It is the most stable quarterback stat we measured, and it matters more in superflex, where you start two.
• Don't chase yards per carry. A back's workload tells you far more about his next month than his average.
• Give it four games. Before that, a single long play can make anyone look hot.

See it in the rankings

Our weekly rankings flag players running hot or cold. REG DOWN means he is scoring well above his chances, and REG UP means well below. Our rankings move flagged players a few spots, and the baseline view shows the order before the move.

Stats: nflverse (CC BY 4.0).