Layered Dirichlet Modeling to Assess the Changing Contributions of MLB Players as they Age

Monnie McGee, Jacob Turner, Bianca Luedeker · 2025

Read Layered Dirichlet Modeling to Assess the Changing Contributions of MLB Players as they Age by Monnie McGee, Jacob Turner, Bianca Luedeker (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-09153.txt.

Introduction

Monnie McGee, Jacob Turner, and Bianca Luedeker model how baseball players contribute as they age in a 2025 paper. Pro careers are short, with most players retired by forty, as speed and strength fade but experience grows. The highest paid players are often over thirty five, partly for mentorship. The authors ask how older players contribute in measurable ways at the plate.

They treat plate appearance outcomes as compositional data across three age groups. Their layered Dirichlet model can test for group differences and pinpoint which outcome type changes most. That framework shows where aging players can be of most use to a team. Readers see a statistical lens on a familiar sports question.

Catalog record

CREATORMonnie McGee, Jacob Turner, Bianca Luedeker
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-09153

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Layered Dirichlet Modeling to Assess the Changing Contributions of MLB Players as they Age Monnie McGee, Jacob Turner, Bianca Luedeker Abstract Abstract The productive career of a professional athlete is limited compared to the normal human lifespan. Most professional athletes have retired by age 40. The early retirement age is due to a combination of age-related performance and life considerations. While younger players typically are stronger and faster than their older teammates, older teammates add value to a team due to their experience and perspective. Indeed, the highest–paid major league baseball players are those over the age of 35. These players contribute intangibly to a team through mentorship of younger players; however, their peak athletic performance has likely passed. Given this, it is of interest to learn how more mature players contribute to a team in measurable ways. We examine the distribution of plate appearance outcomes from three different age groups as compositional data, using Layered Dirichlet Modeling (LDM). We develop a hypothesis testing framework to compare the average proportions of outcomes for each component among 3 of more groups. LDM can not only determine evidence for differences among populations, but also pinpoint within which component the largest changes are likely to occur. This framework can determine where players can be of most use as they age. 1 Introduction In professional baseball, the highest paid players are typically those over the age of 35 [ 13 ] . Perhaps some additional pay is due to their popularity; however, it is unlikely that management would continue to pay large sums of money for a player unless it is perceived that the player is making a contribution to the overall winning percentage of the team. Some of the contribution is likely intangible, such as leadership and mentorship of younger players. Other contributions directly affect game outcomes. For example, there is evidence older players trade off their earlier ability to reach base via foot speed for an ability to reach base via outcomes that require more patience or fortitude, such as walks and hit by pitch [ 4 ] . In this article, we propose a method to determine differences in the distribution of at-bat outcomes, which we model as a composition of six possible at-bat outcomes. We propose a likelihood ratio test for G>2 groups, and apply the test for MLB baseball players near the beginning, middle, and end of their careers. Our analysis corroborates earlier analyses that show the shift in contributions of older players from more explosive outcomes to outcomes that require patience and experience. It is well-known that physical maturation affects athletic performance [ 17 ] , and it is further accepted that athletic performance peaks at a certain age, typically in the mid to late 20’s, and then declines [ 22 ] . Physical characteristics such as speed, strength, lung capacity, and flexibility, are harder to maintain as the body ages, and all of these are related to performance in professional baseball and other sports [ 4 ] . Studies have examined the trajectory of performance-related variables as players age, and all come to the same conclusion: by the time an athlete is 40 years old, most of the markers of athleticism have diminished. However, decline in athleticism can be mitigated by other factors. For example, a study of approximately 34,000 athletes in the Swedish Veteran Athletics data base, which covers track and field events over 120 years [ 14 ] , showed that athletes who started when they were older and at a high performance level tended to experience the slowest rate of decline in performance [ 2 ] . For MLB players, early specialization in baseball has been linked to less longevity due to upper extremity injuries [ 6 ] ; however, a recent review of baseball injuries showed that players can return to play with no discernible difference in performance [ 23 ] . Some studies suggest that peak performance, as well as injury prevalence, interact with the type of activity. More explosive activities requiring strength and speed tend to favor younger athletes, while lower impact activities requiring excellent technique tend to favor more mature athletes [ 9 ] . The experience-productivity relationship for baseball, specifically, has been the subject of several articles. For example, [ 15 ] measured a baseball player’s lifetime batting average using a regression equation including variables for age and age-squared for each of the years that a batter remained in the major leagues. The model also included an error term which consisted of a variable for talent, which is time-invariant, and a time-varying individual performance term [ 15 ] . The aim was to produce a career batting average for each player that was adjusted for age, talent, and experience. This model considered only batting average, and not individual at-bat outcomes. Hakes and Turner (2009) examine the relationship between salary of MLB players and their productivity as measured by on-base percentage plus slugging average (OPS) for 1985-2005. They divide MLB players into quintiles based on the indexed OPS level of a player’s third-best season of more than 130 at bats. Instead of age as a predictor in their model, they use years of experience. Bradbury (2008) defines athleticism for both hitters and pitchers. For hitters, athleticism is defined by a linear regression model with variables for batting average, walks per-plate appearance, double plus triples per at bat, home runs per at bat, on-base percentage, on-base plus slugging percentage, slugging average, and adjusted linear weights. Linear weights is a measure which is a weighted combination of singles, doubles, triples, home runs, walks, hit-by-pitch, stolen bases, caught stealing, and at-bat hits [ 25 ] . The goal is to measure the age at which players peak in their performance according to these measures. Commensurate with most research on performance and aging, the peak age is typically between 27 and 29 for all measures of hitting performance [ 4 ] . Most of the previous work has been to examine age at peak performance, and most of that work used linear regression and one selected measure of performance to determine that peak. We seek to examine the contribution of all offensive plate appearance outcomes of baseball within the context of a game. We treat each outcome as a component of a discrete distribution. The statistical model we utilize to determine changes in age-related performance is based on the nested Dirichlet distribution (NDD) [ 21 ] . We show that there is evidence that the proportions of plate appearance offensive outcomes change for different age groups of batters, and we further show exactly where the largest discrepancies in the outcomes lie. This ability to determine the component with the most impact sets our methodology apart from others. This paper is structured as follows. In Section 2, we introduce the baseball outcome data and explore it with descriptive statistics and graphics. We show that plate appearance outcomes can be considered as individual components that make up a hitter’s performance. To model these components as a whole, we introduce the nested Dirichlet distribution (NDD) in Section 3 and discuss estimation of its parameters. In section 4, we outline a hypothesis test to examine the similarity of

Read the complete text