Abstract

Fantasy managers commonly assume that player production fades in the second half of the season because of cumulative injuries, weather, defensive adjustments, and playoff-related rest. Using weekly fantasy points per game (FP/G) data spanning 2014–2025 for quarterbacks (QB), running backs (RB), tight ends (TE), and wide receivers (WR), we tested whether early-season averages (Weeks 1–9) differ significantly from late-season averages (Weeks 10–18). We applied one-way ANOVA and Tukey HSD post-hoc tests within and across era segments. Results show position-specific patterns rather than a universal decline: QBs exhibit the clearest late-season softening, while RBs and TEs remain statistically stable. These findings refine conventional late-season discounting strategies.

Introduction and Related Work

The belief that fantasy production declines as the season progresses is widespread. Prior published analyses have examined related questions. FantasyLabs (2018) reported that average Actual Points scored by fantasy-relevant players declined during the final five games of the season, attributing the drop primarily to physical attrition and differential playoff motivation.

An earlier study by 4for4 (2010) split seasons into Early (Weeks 1–8) and Late (Weeks 9–17) periods and classified players as “Dazzling Finishers,” “Fading Heroes,” or “Steady Eddies.” That work found that late-season surges or fades carried limited predictive value for the following season.

Footballguys’ “Regression Alert” series and PFF analyses of players who started hot and finished cold further document regression and health/game-script effects.

Our study extends these lines of inquiry with formal ANOVA testing on a continuous 12-year sample, stratified by position and scoring era.

Methods

Dataset: Weekly positional FP/G values from 2014–2025.

Era segments: 2014–2017, 2018–2021, 2022–2025.

Comparison windows: Weeks 1–9 versus Weeks 10–18.

Primary metric: Mean weekly FP/G (secondary: % of TOP normalization).

Statistical tests: One-way ANOVA + Tukey HSD (α = 0.05).

Results

Figure 1. Weekly FP/G Average as Percentage of Top Performer by Position (2014–2025)

Figure 1 legend: Line chart of positional averages expressed as a percentage of that week’s top scorer. The vertical dotted line marks the early/late split (after Week 9). QB and overall W AVG show progressive late-season erosion; RB and TE remain comparatively flat.

Figure 2a. QB Mean Weekly FP/G by Era Segment

Figure 2a legend: Absolute mean FP/G for quarterbacks across the three era segments. Clear downward drift after Week 9 is visible in the 2014–2017 and 2018–2021 cohorts; the 2022–2025 cohort is more stable but still shows modest late softening.

Figure 2b. RB Mean Weekly FP/G by Era Segment (Contrast Case)

Figure 2b legend: Running-back means by the same-era segments. Unlike QBs, RBs display no systematic late-season decline; the lines remain essentially flat across the full 17- to 18-week window.

Figure 3. Early vs Late Season Mean FP/G by Position (with ANOVA Annotations)

Figure 3 legend: Side-by-side bar comparison of early-season (Weeks 1–9) versus late-season (Weeks 10–18) means. Annotated p-values from one-way ANOVA: QB shows a significant decline; RB (p = 0.341) and TE (p = 0.247) are non-significant; selected WR ranking-sector contrasts reach p ≈ 0.03.

Figure 4. Pooled Early vs Late Season Means with Standard Error

Figure 4 legend: Bar chart of the pooled early- versus late-season means (± standard error of the mean) drawn from the study’s primary comparison pool. The late-season mean is lower and the error bar slightly wider, consistent with increased variance.

Discussion

The visual evidence confirms that a blanket “everything declines after Week 9” narrative is incorrect. Running backs and tight ends remain statistically stable (Figures 2b and 3). Quarterbacks exhibit the most reliable late-season softening (Figures 1, 2a, 3, and 4), consistent with weather, defensive adjustments, and rest decisions. Wide receivers occupy an intermediate position.

These patterns align with FantasyLabs’ earlier observation of late-season point decline and 4for4’s finding that intra-season trajectories have limited year-to-year stickiness. Managers should therefore apply position-specific discounts: trust RBs and TEs deeper into the year, exercise greater caution with QBs after the heavy bye-week stretch, and evaluate WRs in ranking-sector context.

Conclusion

Seasonal FP/G trajectories are position-dependent. A multi-year examination of 2014–2025 data demonstrates that progressive decline is clearest for quarterbacks, modest or context-dependent for wide receivers, and unsupported for running backs and tight ends. Fantasy strategy—roster construction, waiver prioritization, and trade valuation—should incorporate these differentiated, data-driven patterns.

References (Selected)

• FantasyLabs (2018). How NFL Teams’ Playoff Odds Affect Fantasy Football Production.

• 4for4 (2010). Consistency Factor – Early vs. Late Season Analysis.

• Footballguys Regression Alert series.

• PFF analyses of hot-start / cold-finish players.

All primary data and statistical outputs are drawn from the 2014–2025 Counts-Over-Median Metrics study (from Science of Fantasy Football - Unpublished Data March 2026).