Content area
Full Text
Abstract
The 2016 NBA Finals is considered one of the best playoff series in recent history, with the Cleveland Cavaliers winning the title against the Golden State Warriors. This result is regarded as a major upset as the 2016 Warriors finished the regular season with an NBA record 73 wins. To analyze the factors that played into the Cavaliers' victory, we develop a discrete-event simulation for basketball games based on publicly available advanced basketball statistics. The model simulates games possession-by-possession and is generalizable to any two teams with appropriate data collection. Model outputs were validated using real game data from the 2016 playoffs. Factors included in the simulation analysis include the Game 5 suspension of Draymond Green, as well as regular season versus playoff performance. Our analysis shows that Cavaliers significantly improved in the postseason while the Warriors postseason performance was worse than during the regular season. Possible reasons for these changes include player fatigue, coaching strategy, and average strength of opponent.
Keywords
Basketball; Analytics; Simulation; Sports; NBA
1.Introduction
The 2016 National Basketball Association (NBA) Finals is regarded as not only one of the best playoff series in recent history, but also as an incredibly unlikely occurrence. Experts from across the basketball world expected the Golden State Warriors (GSW) to defeat the Cleveland Cavaliers (CLE) and many expected it to not be close [1,2]. Thus, when CLE won the series in Game 7 on the back of a heroic performance by LeBron James, many were left asking what happened to lead to the unlikely outcome. Two popular theories were that the GSW players were tired after battling all season to earn an NBA record 73 wins, and that CLE did not put in as much effort in the regular season, thus the team was better than their regular-season statistics indicated. The suspension of GSW star player Draymond Green for a pivotal Game 5 of the series due to a flagrant foul committed in Game 4 was also suspected to have played a significant role.
In this paper we discuss our modeling approach for a simulation model of basketball games that uses individual player statistics as inputs and outputs realistic game data. Next, we use the model to analyze the 2016 Finals. Specifically, we consider...