Fool’s Gold: Can NFL Statistics Spot Overrated Teams and Predict Playoff Upsets?

My paper investigates whether regular-season statistics can identify NFL teams with misleading win totals ("fool's gold") and predict playoff upsets using data from 320 team-seasons across the 2016–2025 seasons. Using predictive models built in R, the study finds that point differential is the single strongest predictor of regular-season win totals, accounting for nearly 80% of win variance. Teams classified as "fool's gold" (those significantly overachieving relative to their point differential) largely relied on exceptional performance in one-score games; these overachievers were roughly 12% more likely to lose their opening playoff game than other postseason teams. However, regular-season metrics offered limited utility in predicting specific playoff upsets, where offensive efficiency was the only significant contributing factor and single-game variance dominated. Ultimately, while regular-season statistics effectively highlight over- and underperforming teams susceptible to regression, they carry weak predictive power for individual playoff outcomes.