NFL Red Flag Games: Peak Performance or Survival of the Fit?

NFL scheduling stress is a hidden crisis. Since 2020, Red Flag games involving short rest plus travel spiked 116%. Using Hex, we prove these cost 0.35 EPA and hit certain positions hardest.

Comment

Inspiration

This project was inspired by a central research question:
How has the cumulative scheduling stressor of Red Flag games impacted team, position-group, and player performance across the league?

As the NFL expanded prime-time windows and optimized schedules for broadcasting, we observed growing concern around short rest weeks and long-distance travel. These factors were typically analyzed in isolation, but we were motivated by the intuition that the intersection of short rest combined with significant time-zone travel creates a compounded stressor that is both underexplored and potentially inequitable. Framing this cumulative burden as a Red Flag game gave us a way to systematically study fatigue, performance, and fairness across the league.

What it does

Our project identifies, quantifies, and analyzes Red Flag games—regular-season NFL games featuring both short rest ((\leq 6) days) and long-distance travel ((\geq 2) time zones). Using this framework, we measure:

  • League-wide exposure trends over time
  • Team-, position-group-, and player-level inequities in exposure
  • Performance impacts associated with cumulative Red Flag exposure
  • Vulnerable cohorts most at risk of measurable performance decline

The analysis reveals that while Red Flag games remain a small fraction of total games, their frequency and concentration have increased significantly, leading to measurable performance penalties and structural inequities—particularly for West Coast teams and specific position groups.

How we built it

We built the project in four stages:

  1. Conceptual framing
    We defined Red Flag games to isolate compounded scheduling stress, rather than treating rest and travel as independent variables.

  2. Dataset creation
    We constructed a multi-year dataset combining NFL schedules, rest-day calculations, stadium geography, and player participation and performance metrics (EPA-based proxies).

  3. Analysis and modeling
    We applied:

    • Trend analysis (pre- vs. post-2020 CBA)
    • Inequality metrics, including the Gini Coefficient
    • K-means clustering to identify high-risk teams, position groups, and players
    • Longitudinal modeling to estimate cumulative performance penalties

  4. Policy translation
    We translated analytical findings into actionable scheduling recommendations focused on equity, feasibility, and player safety.

Challenges we ran into

  • Isolating cumulative effects: Disentangling the combined impact of rest and travel from confounding variables such as opponent strength, injuries, and game context required conservative modeling choices.
  • Data constraints: The absence of public recovery and health data forced us to rely on performance degradation as a proxy for fatigue.
  • Defining fairness: Competitive equity is not synonymous with equal exposure; balancing geographic realities with league operations was a conceptual challenge.
  • Bridging analysis and policy: Recommendations had to be impactful while remaining realistic within the NFL’s broadcast-driven scheduling ecosystem.

Accomplishments that we're proud of

  • Defining and operationalizing Red Flag games as a novel cumulative scheduling stressor
  • Quantifying league-wide inequity with a 2024 Gini Coefficient of 0.774 for exposure
  • Identifying 105 players (8.6%) as High-Risk Vulnerable with measurable performance decline
  • Demonstrating a statistically meaningful performance penalty of up to (-0.351) EPA per play
  • Producing policy recommendations that could reduce harm while affecting less than 2% of total games

What we learned

We learned that cumulative workload stress is both measurable and highly concentrated. A small subset of teams and players absorbs a disproportionate share of scheduling burden, and those exposures translate into real performance consequences. We also learned that geography plays a systemic role in competitive equity and that unsupervised learning methods are powerful tools for uncovering hidden vulnerability in sports data.

Most importantly, we learned that structural issues in scheduling persist not because they are unavoidable—but because they are rarely measured holistically.

What's next for Untitled

Next, we plan to:

  • Integrate injury and recovery indicators as data availability improves
  • Simulate alternative scheduling algorithms to minimize Red Flag exposure
  • Expand analysis to international games and additional rest/travel thresholds
  • Evaluate how roster rotation and organizational strategies mitigate cumulative stress

Ultimately, our goal is to move from diagnosis to implementation by helping design an NFL schedule that better balances competitive equity, player health, and modern broadcast demands.

Built With

Share this project:

Updates