Workshop
Evaluating team strength from scratch
Darina Goldin
Director Data Science, Bayes Esports
Dr. Darina Goldin is Director Data Science at Bayes esports where for the longest time she was leading the betting team. Darina is recognized by her peers as a leading practitioner with 6+ years of experience and validated expertise in Data Analytics and Data Science in Esports. She has completed a PhD in Control Science at the Technical University of Berlin.
Session description
Evaluating team strength is a cornerstone of sports analytics, and several algorithms exist for this task. In this workshop we will first deal with the annoying but crucial task of getting the data - connecting to an API, authenticating, downloading, and rewriting the results. We will then throw the various python implementations of the rating algorithms at the data we have collected and compare the results.
By the end of the workshop, you will understand how to estimate team strength in sports and make win predictions using python implementations of different rating algorithms.
Structure:
Introduction to team ratings and predictions
Write our own implementation of Elo and apply it to the data
Create ratings with Glicko and Trueskill using existing packages
Compare results, experiment, and discuss how the ratings can be improved
Requirements:
Basic understanding of Python
Have a virtual environment with Jupyter
Get the code from this repo: https://github.com/drdarina/datalift_workshop and follow the Setup instructions
Please test if you can run the imports in the workbook before the workshop