The Madness of Multiple Entries in March Madness
Presented at:
- Talk, INFORMS Annual Meeting, Seattle, USA (October 20-23, 2024)
- Poster, ICERM: Discrete Optimization, Brown University, Providence, USA (August 26-30, 2024)
- Poster, Economics and Computation, Yale University, New Haven, USA (July 08-11, 2024)
- Poster, NYC Operations Day, Columbia University, New York, USA (May 10, 2024)
- Talk, INFORMS Annual Meeting, Phoenix, USA (October 15-18, 2023)
- Talk, 29th International Conference on Principles and Practice of Constraint Programming, University of Toronto, Toronto, Canada (August 27-31, 2023)
Abstract
This study addresses the challenge of optimal portfolio allocation in a classical betting setting, assuming the bettor seeks to maximize log-utility. While the Kelly Criterion is well-established as the optimal allocation for a single bet, questions remain open regarding how to proceed when considering multiple bets, simultaneously. Building on prior literature, we present theoretical results on optimal parlay construction and propose a scalable logic-based Benders decomposition algorithm. We evaluate the algorithm’s performance on both synthetic data and real-world sports betting instances, demonstrating its potential to enhance parlay strategies.