Predicting Formula 1 Race Outcomes: Decomposing the Roles of Drivers and Constructors through Linear Modeling

TL;DR

This paper introduces a novel linear modeling approach to disentangle and quantify the individual impacts of drivers and constructors on Formula 1 race outcomes, providing clearer insights into their respective contributions.

Contribution

It extends the RAPM methodology with time-decayed ridge regression and LOESS smoothing to isolate driver and constructor effects over a decade of F1 races.

Findings

Constructors explain 64% of race outcome variance.

Constructor impact increased in rank-agnostic cohorts.

Driver and constructor contributions vary over time.

Abstract

Formula 1 performance is a combination of the car's ability and the driver's ability. While a given race or season can tell you how well a car and driver performed jointly, isolating the individual impact of the driver and constructor remains challenging. This paper extends a Regularized Adjusted Plus Minus (RAPM) methodology (Sill 2010), commonly used in basketball and hockey, to parse out individual driver and constructor impact. It employs a time-decayed ridge regression with LOESS (Jacoby 2000) smoothing to predict race results for the Hybrid Engine Era (2014 - 2024). By measuring the constructor and driver coefficients over time, we measure the relative individual impact of driver and constructor throughout the period. Results show that constructors explain 64.0% of the variance in race outcomes in the Hybrid Engine Era. Additionally, constructors have increased importance in benchmarked rank-agnostic cohorts (e.g., Top 10 points finishers) and decreased importance in qualifying. By decomposing performance into individual driver and constructor metrics, we create a robust framework for inter-constructor driver comparisons that the Formula 1 points system obfuscates. Our work enhances the understanding of driver and constructor contributions to race success, offering valuable insights for strategic decision-making in Formula 1.