Rescheduling the NBA regular season via Integer Programming

GARCÍA ARAMOUNI, NICOLÁS; MIRANDA BRONT, JUAN JOSÉ

Santiago de Chile

Conferencia; 23rd Conference of the International Federation of Operational Research Societies; 2023

IFORS

The COVID-19 pandemic generated disruptions across multiple sectors; in particular, in the sports industry as many leagues had to re-adapt their competitions as a consequence of lock-downs, travel restrictions and other implemented safety measures. Even when activities started to resume, match suspensions continued throughout the different leagues. Dealing effectively with these unexpected events is extremely relevant regarding both sports and economic aspects of the competitions. In this paper, we propose a framework building upon Integer Programming models to systematically reschedule suspended games and generate a contingency fixture accounting for relevant operational constraints. For this purpose, we consider and compare two reactive strategies: a constrained approach, where disrupted games are rescheduled into feasible dates without modifying the original fixture, and a flexible framework, allowed to introduce a controlled number of small perturbations to the original fixture to improve the solution. Using the 2020-21 NBA season as a benchmark, we compare different scheduling approaches under different objective functions and discuss different characteristics of the solutions obtained. The approach produces good quality schedules, and the framework has potential to be applied in practice.