Fantasy Football Week 3: Breakout Model Picks

Sep 23, 2026

IMAGN IMAGES via Reuters Connect

With two weeks of data, it’s time to look at the breakout WR model for the first time this season. Let’s give some details on how this works.

About the Breakout Receiver Model

The table below shows pre-snap field position on the x-axis. That field position ranges from backed up in your own end zone on the left side to the opponent’s goal line on the right side. And then air yards are on the y-axis. The numbers in the table show the average number of half-PPR fantasy points that a target in that specific bucket earned in 2025. So, a throw behind the line of scrimmage backed up in your own end zone scored 0.8 fantasy points on average. But that same throw near the opponent’s end zone averaged 1.7 points.

So, I made a machine learning model (XGBoost for my fellow nerds) that looked at every target from 2021 to 2025. It uses pre-snap field position and air yards to predict how many fantasy points a given target is expected to earn. That ultimately made what I’m calling a prediction surface, which is pictured below. It looks an awful lot like the table above, but it smooths out some of the fluctuations and provides a prediction for every possible combination, even those that didn’t occur often in 2025.

Now that we know how many fantasy points every target *should* earn, we can compare those expected fantasy points to what actually happened. The graph below does that for every player through two weeks of the season. Players far below the dotted line are underperforming their expected points and are potential buy-low candidates. And players above the line are overperforming their usage and are potential sell-high candidates. That lets us ask important questions like “who the hell is Ryan Miller?”

Before turning to more players, I didn’t want to leave out the RBs. It will surprise no one that carries closer to the goal line are more valuable than those on other parts of the field. That’s why we often say that carries between the 20s don’t really matter for fantasy football. The graph below shows pre-snap field position on the x-axis and average fantasy points per carry on the y-axis. A run 10 or fewer yards from the opponent’s end zone is four times as valuable as essentially any other carry.

Just like for WRs and TEs, I built a machine learning model to calculate expected points per carry. It mostly uses field position but also has a small adjustment for down, where 3rd down carries are slightly more valuable and 4th down carries are slightly less valuable.