By Karsten Maurer (Miami University)
Abstract
A fantasy sport is a game played using data from a professional sport. Participants competitively draft the names of professional athletes who they believe will perform well - as measured by their real world statistics - to be added to the roster of an imaginary team. Drafting a fantasy sports roster can incorporate strategic planning based on predictive modeling; where accurate forecasts of performance can give a team a better chance of picking better players. Students of our Statistical Learning course formed groups that each prepared a drafting strategy, constructed and validated predictive models, conducted an in-class fantasy basketball draft, and finally competed in a live-action simulated tournament. The project provided a quality synthesis of all supervised learning methods covered in the course. The combination of a popular subject matter and low-stakes competition with classmates also motivated some students to expand their learning well beyond the requirements of the course.