Fantasy Football Projections: How to Use Them and What They Cannot Tell You
Every platform, app, and analyst produces weekly projections: Player A is projected for 14.3 points, Player B for 12.8. The temptation is to start Player A over Player B every time. But projections are estimates with significant uncertainty ranges, and treating them as precise predictions leads to poor decisions.
Projection Quick Reference
| Situation | How to use projections | What to override with |
|---|---|---|
| Two players within 2 points of each other | Effectively a coin flip — do not start/sit based on the point gap alone | Underlying usage stability, news, matchup quality |
| Player had a huge game last week | Projection will spike; check if the usage was real or situational | Was last week’s role change permanent or game-script dependent? |
| Player had a terrible game last week | Projection may underrate him if the bad game was variance | Is his opportunity still intact? Check routes run and target share |
| Game-time scratch or late injury news | Projection is now stale — ignore it | Real-time injury reports always supersede mid-week projections |
| Draft and trade decisions | Season-level projections are much more reliable than weekly | Player A projected 185 pts vs Player B at 175 pts matters more than any single-week comparison |
| New situation (new team, new OC, rookie) | Projections lag new information by 1–2 weeks | Manually check snap data, routes run, and beat reporter notes |
What Projections Actually Measure
A fantasy football projection is a point estimate for what a player is expected to score given a specific set of assumptions: the player’s target share/carries based on recent games, the opponent’s defensive strength against the position, the game’s expected total (from Vegas lines), the player’s injury status and snap rate, and recent efficiency metrics.
What it does not say: that the player will actually score that number (variance is enormous); that the assumptions are correct (a coaching decision could reduce his role this week); that it accounts for unknowns (a weather event, late scratch, in-game momentum shift).
How Accurate Are Projections?
The variance problem: A player projected for 15 points might realistically score anywhere from 5 to 35 in any given week. The standard deviation of a WR’s weekly score is typically 8–12 points from their projection. This means two players projected for 14.5 and 12.5 are so close that the 12.5 player will outscore the 14.5 player roughly 40% of the time just from variance.
Season-level vs week-level: Across a 17-week season, a player projected for 200 points will likely finish closer to 200 than he will to any particular weekly projection. Volume averages out over large samples. Week-to-week, variance dominates. Projections are most useful for drafts and trades; least useful for marginal start/sit decisions.
When to Trust Projections More
Stable underlying usage: A bellcow RB with 22–25 carries per game and a consistent pass-down role has a more reliable projection than a committee back whose carries vary from 8 to 20 based on game script.
Clear matchup data: If a team ranks 30th against wide receivers, that information is real and projections incorporating it are more reliable. When a team ranks 15th with no clear tendency, the matchup adjustment is less meaningful.
Elite players: Players in the top tier at their position have more stable projections because their role is clearer. A QB1 throwing 40+ times per game in a high-scoring offense has a more reliable weekly floor than a streaming QB whose usage depends on game script.
When to Be Skeptical
After a big game: Projections often overadjust. A RB who had 30 carries in Week 6 might see his Week 7 projection spike, but the 30-carry game might have been an outlier. Check the opportunity that generated the big game, not just the score.
After a bad game: A player who scored 3 points due to a fluke may be underrated next week if the underlying usage is intact.
New situations: Projections lag new information. A WR whose slot target share jumped because the incumbent got injured in Week 3 may not have updated projections until Week 4 or 5. Manually checking snap data is more valuable than waiting for projections to catch up.
Late information: Projections are often set mid-week. If a player is listed as questionable Sunday morning, his projection does not automatically update. Real-time information always supersedes a mid-week projection.
How to Use Projections Well
- Use them as a baseline, not a conclusion — identify which players are projected to contribute, then layer in manual judgment
- Discount marginal differences — a 2-point gap is close enough that specific situational factors matter more than the number
- Update for late information — re-evaluate your lineup after Friday injury reports and Sunday morning news regardless of what the weekly projection said Tuesday
- Use season-level projections for draft and trade decisions — much more reliable guidance than single-week comparisons
For the specific sit/start framework that incorporates projections alongside other factors, see: Sit/Start Decision Framework: How to Set Your Lineup Every Week. For how matchup information specifically affects projections, see: How to Evaluate Fantasy Football Matchups and Use Them to Your Advantage.