The interplay of luck and skill in sports is a complicated and highly discussed topic. This paper aims to quantify and interpret the role of luck in various sport leagues. Using Classical Test Theory and the assumption that skill differences between teams become more apparent over the course of a season, the authors estimate luck contributions by comparing the variance of point distributions between actual and simulated results in a "luck-only" scenario.
In order to estimate the luck contribution run Variance_method.ipynb.
At this point the work for climbing and bouldering is highly experimental and cannot be compared to the results of the leagues!
The entropy_method.ipynb is also highly experimental
The preprocessing is already done. The results are scraped in data/data_preparation/scraping_standings. The schedules are scraped in data/data_preparation/scraping_schedules. Some preparation is done in data/data_preparation/preparation. The data needed for Variance_method.ipynb for each league is in data/prepared_data. The schedules for each league and year available are in data/prepared_data/schedules.
Config data for each league can be found in config.yaml.
The Luck_vs_skill_report.pdf contains more information about the theory behind this process.
