Towards a more efficient scheduling of regular NBA games

Abstract

Finding the optimal schedule for the 82 games played by each of its 32 teams in a regular season is a continuing challenge for the National Basketball Association (NBA). We examine the schedules of the regular NBA games in the last 22 seasons including the current one to find possible trends in its key parameters - distance traveled during road games and number of successive (e.g. back-to-back) games played. To reduce operational cost and to keep a level playing field, the total distance traveled and the number of back-to-back games must be kept to a minimal and the same for all teams. Data have shown that teams are unlikely to win twice in back-to-back games. We implement a scheduling algorithm that aims to minimize a cost function that incorporates the key parameters as its variables. For the same season duration, the algorithm has been able to reduce the number of back-to-back games by 24% while increasing the total distance traveled by 25%.