Fantasy Football Quarterback Projections: Key Metrics and Modeling
Quarterback projections sit at the intersection of raw volume statistics and contextual modeling — where pass attempts per game, target distribution, and defensive opponent rankings all fold together into a single expected point total. This page covers the core inputs that drive QB projection models, the mechanical logic connecting raw data to fantasy output, and the decision thresholds that separate a reliable projection from a noisy guess. Getting QB projections right matters more in superflex and two-QB formats, where the position's scarcity is amplified and mispricing a starter carries immediate lineup cost.
Definition and scope
A quarterback projection is a forward-looking estimate of how many fantasy points a specific QB is expected to score in a given game or across a defined period — typically a single week, the remainder of the season, or a full-season preseason outlook. The estimate is built from a combination of historical performance data, opponent-adjusted metrics, situational variables, and scoring-format rules.
The scope is narrower than it sounds. A projection is not a ranking, and treating it as one is one of the more common analytical errors in fantasy football. A projection answers how many points — a ranking answers compared to whom. The distinction matters most at the positional boundary: a QB projected for 22 fantasy points in a half-PPR league is useful only once that number is weighed against replacement-level production. The projection vs. ranking difference is worth understanding before applying any QB number to a roster decision.
How it works
The mechanical pipeline starts with opportunity volume. Passing attempts per game is the single largest predictor of fantasy output for quarterbacks — more so than efficiency metrics like yards per attempt. A QB who throws 40 times in a weak offensive scheme will, on average, outscore a QB who throws 28 times with elite supporting receivers. That volume baseline is then adjusted through four major filters:
-
Opponent pass defense ranking — measured by yards allowed per dropback, defensive DVOA (Football Outsiders publishes position-level DVOA figures), and implied passing-game usage drawn from Vegas game totals and spreads. Vegas lines and their relationship to fantasy projections offer a reliable external signal that isn't subject to the same recency biases as trailing box scores.
-
Game script probability — expected score differential affects how many pass attempts a team actually deploys. A QB on a team projected as a 10-point favorite runs a meaningfully higher risk of being benched in the fourth quarter, reducing counting stats. Models built on in-season versus preseason projection logic weight this factor differently depending on the sample available.
-
Target distribution and pass-catcher health — a QB's floor and ceiling both shift significantly when a WR1 misses practice. The dependency runs in both directions: when Tyreek Hill played all 17 games for the Kansas City Chiefs in the 2021 season, Patrick Mahomes averaged over 300 yards passing per game. Target concentration models track snap count and target share data to identify when a quarterback is losing or gaining a top outlet.
-
Scoring format multipliers — standard, half-PPR, and full-PPR formats don't change QB scoring directly, but they shift positional value relative to flex positions, which affects how aggressively QB projections should be applied in lineup decisions. Scoring format impact on projections covers this in detail.
Common scenarios
Three situations push QB projections away from their baseline most reliably:
High-total games with a poor pass defense — when a Vegas total exceeds 50 points and the opposing defense ranks in the bottom third of the league by DVOA, projection models routinely adjust passing volume upward by 15–20% relative to season average. This is one of the cleaner correlations in the data.
Dome vs. outdoor cold-weather games — wind above 15 mph at game time reduces average passing yards by a measurable margin; the specific impact of weather inputs is detailed at weather impact on fantasy projections. Projection systems that don't incorporate game-time conditions will systematically overestimate production in December road games.
Two-QB and superflex leagues — the entire projected point distribution shifts upward, and floor and ceiling projections become the operative decision tool rather than a single-point estimate. A QB with a high floor and modest ceiling is more valuable in these formats than a boom-bust arm with equivalent expected points.
Decision boundaries
The threshold questions for applying QB projections in live roster decisions come down to three comparisons:
- Starter vs. streamer: a QB's projected output versus the best available free-agent QB at game time. If the gap is under 4 projected points in a single-QB format, matchup becomes the tiebreaker.
- Trade value calibration: season-long projections feed directly into trade value and projection data — a QB with a favorable second-half schedule carries higher rest-of-season value than raw YTD numbers suggest.
- Sample reliability: projections built on fewer than 6 games carry a sample size and projection reliability caveat — variance is high enough that the confidence interval around the point estimate is essentially a range, not a number.
The full projection methodology for quarterbacks, including position-specific model architecture, is documented at quarterback projection methodology. For a broader orientation to how all the inputs connect, the home page offers a structured entry point to the full projection framework.