Analytics for Local Collegiate Baseball League : Improved Statistics and Favorable Factors

Abstract

This project focuses on analytics methods based on traditional, historic statistics gathered for baseball players, as well as team win-loss records within a defined competitor framework. Methodologies for both team-level and player-level analyses were adjusted for the Northwoods League, including the local team Eau Claire Express, using historical data. We hope to be able to provide value to the local community by sharing some of the insights gained. Assessments of individual player batting and pitching strengths were computed, based on statistics developed recently within Major League Baseball; explanation of these metrics is available on sites such as at FanGraphs. Comparisons of these newer metrics are made to historical assessment measures. Summaries of team records were gathered across the most recent four seasons, for 18-20 teams in the league. Various recursive record-updating methods were considered for predictive purposes. The current analysis examines summary statistic values that appear to be most associated with streaks of wins or losses. Methods for modeling streaks by incorporating team statistics and other metrics are examined. This outline to the right tracks the research, data management, and programming used to gather, clean, compute, organize, and evaluate the team and individual player data for the Northwoods League.