Fourth downs decoded : a predictive and causal analysis of player impact and decision making in the NFL

Abstract

This research breaks down fourth downs in the NFL by identifying crucial player positions and evaluating how coaches incorporate different factors in their decision-making process. Using a newly engineered dataset that incorporated detailed player statistics from 2017-2023, we analyze the causal and predictive factors influencing fourth down decisions and outcomes. Our Generalized Inverse Mills Ratio reveals coaches factor in unobserved variables during fourth down decisions that can predict the likelihood of a fourth down conversion. However certain player archetypes, such as quarterbacks that excel in short passes, are not properly evaluated. We find significant impacts from other specific player positions, with offensive linemen showing strong joint significance on conversion but not on coaching decisions. The findings provide actionable insights for NFL coaches to identify key signals in fourth down situations and optimize player selection and team build for these critical plays.