This study examines whether head coach scheduled pay, football program investment, and an in-season performance indicator are associated with winning percentage in NCAA Division I Football Bowl Subdivision (FBS) football, and whether conference type moderates the program investment–performance relationship. The study additionally demonstrates how scalable, AI-assisted data integration can support reproducible sport management research.
Using 1,373 team-season observations across thirteen competitive seasons (2010–2019 and 2022–2024), the study implements a reproducible data integration workflow that programmatically retrieves multi-level game data via Python APIs, engineers features and harmonizes identifiers in R, and stores integrated records in a relational SQLite database. Winning percentage serves as the dependent variable. Ordinary least squares (OLS) regression models evaluate the direct effects of coach pay, program spending, and season-average scoring margin, as well as a spending × conference interaction term to test moderation.
Regression results indicate that coach pay and program spending are positively associated with winning percentage in separate baseline models, while scoring margin exhibits a substantially stronger effect. In the full specification including all predictors simultaneously, the effects of financial inputs attenuate substantially and are weak or non-significant, suggesting that financial resources operate primarily through improvements in on-field performance rather than through a direct independent pathway. The interaction between spending and conference type is not statistically significant.
By integrating theory-driven hypotheses with a reproducible, AI-assisted data workflow spanning over a decade of FBS seasons, this study contributes to the intersection of sport finance, analytics, and data-driven decision-making in intercollegiate athletics, while offering a replicable methodological framework for future multi-source sport management research.
