Analytics for Summer Collegiate Baseball : Connecting Individual and Team Results

Abstract

A prior study on baseball analytics for the Northwoods League summarized the available components needed to calculate a "Win Above Replacement" (WAR) metric for individual players, as well as identified missing and potential replacement measures in this league. Some of the inputs were found to be unobtainable due to lack of technical equipment in the league, but research into the origins of the measure and historic records allowed us to identify some suitable substitute measures. Obtaining these measures is possible due solely to numerous programming and data-gleaning achievements. These include code to pull and summarize play-by-play information from the original online text source; back-computing the physical locations of defensive plays from an image of the playing field; and creating a system both to extract and to compute player and team metrics beyond those automatically provided for the Northwoods League data. Summary of these methods will be included in the presentation, along with a discussion of the structure of the usable data, including play-by-play output, individual-player, and league-level summaries. Additionally, progress towards a WAR-analogy estimation (using a model connecting team performance and player appearance) will be included.