Under pressure: Integrating machine learning to quantify clutch moments in American football.

This article presents a novel pressure index framework for quantifying situational pressure in American football by integrating key game-specific variables such as time remaining, score differential, field position, and down progression. It introduces two main indices: the Composite Offensive Pressure Index (PIO3) and the Defensive Pressure Index (PID), which dynamically measure offensive and defensive pressure during a game. Using play-by-play data from the 2023 NFL season, the study validates these indices through machine learning models—Logistic Regression, Random Forest, and Gradient Boosting—with Gradient Boosting achieving the highest predictive accuracy (AUC = 0.944) in forecasting drive success defined by touchdowns. The framework offers interpretable insights into team performance under pressure, highlights differences in how NFL teams manage high-pressure situations, and suggests applications in coaching, performance evaluation, and broader sports analytics research.