Fantasy Football Projections: How to Read Them, Where They Come From, and Their Limits
A weekly projection that says a running back is projected for 14.2 points feels precise — but that number carries far more uncertainty than two decimal places suggest. Knowing what projections are based on and where they fail is more useful than knowing which site’s projections are “most accurate.”
Fantasy Projections Quick Reference
| Input type | Reliability | How to use it | What it misses |
|---|---|---|---|
| Historical player data | Moderate — past production is baseline, not guarantee | Use as starting floor for expected volume | Doesn’t capture recent role changes; slow to update |
| Depth chart / snap share | High — most important role input | Prioritize; confirms or contradicts historical data | Doesn’t capture intra-game usage patterns |
| Opponent defense | Low-moderate — varies game-to-game | Use as tiebreaker, not primary signal | Individual matchups override unit averages |
| Vegas line / game total | High for game script; moderate for player-level | Use for game flow predictions; informs whether team will pass or run | Volume projections require additional step: who gets those targets? |
| Touchdown projections | Low — highest-variance component | Treat as noise in single-week projections | Can’t predict which red zone trip converts; RZ targets matter more |
| Injury-adjacent players (Q tag) | Very low — projection system must guess at playing time | Ignore automated projections; manually assess probability | Up-to-date injury news is not captured in generated projections |
How Projections Are Built
Historical player data. Statistical models use a player’s prior season output, recent game production, and career trends as a baseline. A WR who averaged 8 targets per game over the prior three seasons will have that volume as the starting point for projections this year.
Depth chart and snap share. Who is the starter? How many snaps are they expected to play? These inputs matter more than any individual game stat — a player who was a WR3 last year but steps into a WR1 role will see significantly higher projected targets.
Opponent defense. Projections adjust for matchup — teams that historically give up more fantasy points to a given position will inflate projections for players facing that defense. This adjustment is one of the weaker inputs: defensive strength against a position varies game to game, and some defenses that look weak early in the season tighten up.
Vegas lines and game totals. The implied total points in a game (the over/under) is a major input for many projection systems. A team playing in a high-scoring game (50+ total) will have different projected volume than a team playing in a projected 37-point game. Teams playing in high-scoring games tend to throw more, which benefits passing game assets.
What Projections Are and Are Not
Projections are probability estimates, not predictions. A projection of 14.2 points is not saying “this player will score 14.2 points.” It is saying that based on available information, 14.2 is the central expected value — but the player might score 4, might score 32, and anything in between is plausible. The distribution around that number is typically wide.
Consensus projections average out information. The “consensus” projection from aggregating multiple sites tends to be more accurate than any individual projection, because averaging reduces the noise introduced by any single analyst’s assumptions. But consensus also eliminates extreme views — the person who projects 22 points based on insider knowledge about a player’s expanded role would be averaged down toward 14.
Projections do not account for breaking news. A projection generated Monday morning does not know that a starter was scratched during Saturday warmups. Injury news, late inactives, and weather changes are not incorporated automatically.
Where Projections Consistently Fail
Snap and target volatility. Projections assume a reasonably stable role. When a team’s offensive coordinator changes the game plan week to week, or when a player’s snap share is inconsistent, even well-built projections miss badly.
Touchdowns. Touchdowns are the highest-variance component of a projection. Whether a team scores three times from inside the 10-yard line or once is not reliably predictable — the projection includes an expected value, but a player with “0.6 expected touchdowns” might have 2 or might have 0.
Injury-adjacent situations. When a player is listed as questionable, projections handle this poorly. Some systems project the player as if they are healthy; others project a reduced role. Neither approach is reliably correct.
How to Use Projections Effectively
Use them as a starting point, not a final answer. Projections help establish a baseline expectation. They are most useful for identifying which players are in high-volume situations and which are in limited ones.
Prioritize volume metrics over projected points. A player projected for 9 targets is easier to trust than a player projected for 14 points built on 5 targets and a projected touchdown. Volume is more repeatable than scoring predictions.
Be skeptical of large deviations from consensus. If one projection system has a player at 22 points and the consensus is 12, that is not necessarily a signal to start the player — it may signal that the system has an unusual assumption that will not pan out.
For how projections connect to lineup decisions throughout the week, see: Fantasy Football Weekly Strategy: How to Make the Best Decisions Every Week of the Season. For how game totals and game script factor into who benefits, see: Fantasy Football Game Flow: How Game Script and Scoring Situations Affect Player Production.