Optimizing Roster Composition in Major League Baseball

Open Access
- Author:
- Kroboth, Kyle
- Area of Honors:
- Statistics
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Andrew John Wiesner, Thesis Supervisor
Matthew D Beckman, Thesis Honors Advisor - Keywords:
- sports
analytics
baseball
market size
regression
statistics
linear - Abstract:
- Major League Baseball (MLB) teams are constantly working to gain an edge on their opponents. The MLB is the only major North American sports league without a salary cap, and as a result, one area that fans heavily scrutinize is teams’ payrolls and how they are building out their rosters. However, some teams are top spenders and fail, while some bargain hunting teams find themselves in regular playoff contention. This thesis aims to determine the blueprint for a successful baseball team based on payroll, salary distribution, service time, and more. Success will be dictated by winning percentage, through a multiple linear regression model, and by playoff appearance odds, through a logistic regression. The models were also run with a subset of data strictly including “small market” teams. After testing assumptions, running regressions, and selecting models, all models were found to show a relationship between team payroll and how many players take up half of that team’s payroll with both winning percentage and playoff odds. Surprisingly, payroll was not determined to correlate with success when limiting the models to small market teams, leaving only the number of players to 50% of payroll as a predictor. Overall, it was found that most of a team’s success is derived from factors outside of roster composition, and even in the top model, only roughly 27% of variation in winning percentage could be described by roster composition predictors.