# Causal effect analysis of serving performance using double machine learning

## TL;DR

This study uses machine learning to analyze how serve performance affects match outcomes in men's professional tennis.

## Contribution

The paper introduces a causal analysis of serve-related indicators using double machine learning in tennis performance evaluation.

## Key findings

- Ace rate has a modest positive causal association with winning probability.
- First serve win rate and first serve in rate have small, context-dependent impacts.
- Double fault rate effects are limited and statistically robust across models.

## Abstract

Serving performance is widely recognized as a critical factor influencing match outcomes in professional tennis. To evaluate its contribution to winning probability, this study analyzes ATP men’s singles matches (2013–2024) and estimates the causal effects of four serve-related indicators: ace rate, first serve win rate, first serve in rate, and double fault rate.Results indicate that the ace rate shows a modest positive causal association rather than a uniformly negative one, while first serve win rate and first serve in rate exhibit context-dependent but statistically small impacts, and the double fault rate effects remain limited.These effects, although moderate in magnitude, remain statistically robust across multiple model specifications.The findings highlight the importance of adapting serve strategies across surfaces, ranking groups, and tournament levels.This study focuses exclusively on ATP men’s singles data, and future research should validate these causal relationships in WTA and mixed competitions to enhance generalizability.