Sports Analytics: The Challenge of Soccer

The field of player analytics is challenging. Or maybe it is just a mess. Many smart folks are working in the area, and a lot of cool stuff has been created over time. But the field ranges from analysts working in front offices to amateurs in the proverbial basement and even an occasional academic. The problems are usually sport-specific, and the results are frequently proprietary. There are also different types of player analytics. Some statistics are geared to rating players, and other models are designed to forecast future performance. The end result is a fragmented field that may lack user-friendliness. The paper linked below is from a research project to create a flexible and generalizable approach to player analytics. Specifically, the approach is designed to solve several common challenges in player analytics. The first challenge is the issue of "Big Data." The current data environment often includes vast quantities of data related to player outputs and, critically, player activities. This issue is more opportunity than a challenge, as more information is potentially a good problem. The second challenge is relating player performance to team outcomes. Lots of statistics add up a bunch of things to come up with a summary measure. Player Efficiency in basketball and Passer Rating in football are two examples. A dilemma with these multi-attribute models is that it's tough to figure out how to weight different player activities. The third challenge is something goes way beyond sports. Analytics is something that usually informs rather than drives decision making. Whether constructing a simple sports statistic or a generalizable method for performing player analytics, it is critical to understand how the analytics product will be used. Human decision-makers are often PhDs in their chosen sport, but they do suffer from cognitive biases. To demonstrate the method, we use data on Fullbacks from Major League Soccer. Nothing magical about this specific sport or position, but it is a low scoring position that engages in lots of defensive and support activities. It is an ideal position to demonstrate the method.

  • Facebook
  • Twitter
  • Instagram
  • iTunes
  • Spotify

© 2020 by Fanalytics

GBS_MarketingAnalyticsCenter_hz_rv_2L.pn