Expected fantasy points

Expected points, week by week

Expected = what his targets and carries should have scoredActual = the box scoreDiff = actual minus expected

How to read it

What expected fantasy points are

Every target is priced by how deep it travelled and where on the field it was thrown. Every carry is priced by the yard line and whether it was short yardage. Every pass attempt is priced the same way for the quarterback, 4 points a passing score. Plays in garbage time (17 up or down inside the last 8 minutes, 25 in the second half) carry their own, lower price. The price is the average points that kind of play produced across 2023 to 2025. Add up a player's plays and you get what his usage should have scored.

Why it matters more than points

Tested on 2025 with the model fit on 2023 and 2024 only, expected points predicted the next week's score better than the score itself did: backs r .40 against .31, receivers .25 against .20, tight ends .24 against .14. Players who beat their expected by six or more scored their expected the next week, not their box score. The big green diffs are the sell-highs. The big red ones are the buy-lows.

What it is not

It is not a skill grade. A player who runs hot all year is usually good, and a player who runs cold for a month is sometimes bad. One week is one week. The season tab is where the diff starts to mean something. For quarterbacks it barely beats the box score (r .20 against .19), so read the QB tab as usage, not as a call.

Where the numbers come from

Play by play and the weekly box score from nflverse. The ffverse column is the open-source expected points model from the same community, printed as a cross-check. Our model agreed with it at r = 0.98 on 2025 without ever seeing it. Some paid services chart whether a target was catchable; play by play does not know that, so we do not pretend to.