Functional Data Analysis of Aging Curves in Sports

TL;DR

This paper applies functional data analysis to study aging curves in sports, revealing distinct aging patterns and differences in player performance over time in NBA and MLB data.

Contribution

It introduces FDA techniques to aging curve analysis in sports, providing more flexible and detailed insights than previous rudimentary methods.

Findings

Identified three distinct aging patterns among NBA players.

Showed differences in aging curves between power hitters and non-power hitters.

Demonstrated independence of aging patterns from player position.

Abstract

It is well known that athletic and physical condition is affected by age. Plotting an individual athlete's performance against age creates a graph commonly called the player's aging curve. Despite the obvious interest to coaches and managers, the analysis of aging curves so far has used fairly rudimentary techniques. In this paper, we introduce functional data analysis (FDA) to the study of aging curves in sports and argue that it is both more general and more flexible compared to the methods that have previously been used. We also illustrate the rich analysis that is possible by analyzing data for NBA and MLB players. In the analysis of MLB data, we use functional principal components analysis (fPCA) to perform functional hypothesis testing and show differences in aging curves between potential power hitters and potential non-power hitters. The analysis of aging curves in NBA players illustrates the use of the PACE method. We show that there are three distinct aging patterns among NBA players and that player scoring ability differs across the patterns. We also show that aging pattern is independent of position.