This is a review of the player tracking data and network model research conducted by Brian Skinner and Stephen J. Guy.
Since the 2013/2014 season, all NBA arenas have installed a system of cameras and tracking software. These systems provide a wealth of information that can then be analyzed. This information includes both quantifiable and non-quantifiable skills. Quantifiable skills include points, rebounds, assists, steals, and blocks. Non-quantifiable skills include high quality (non-assist) passes, setting up good screens or effective rotation on defense. This tracking information is not currently available to the public. If, or when, it does become available these researchers suggest creating a network model to analyze this information. It is possible to create a network model of a basketball offense that relates players’ skills to the team’s success at running different plays. If the tracking information were inputted into such a model the output would give analysts and teams the capability to predict the ability and effectiveness of any 5-man lineups.
Traditionally, analysts spent a considerable amount of time watching players perform in order to evaluate different aspects of their performance, with an emphasis on the non-quantifiable aspects that could not be seen on the score sheet. This combination of tracking data and network model would provide analysts with unbiased information of the different players’ skills levels and how they would work together as a team. Every aspect of the game would be open to evaluation. Analysts could learn the skill levels of NBA players and use this to make predictions regarding the team’s performance in the upcoming season.
Using the player tracking data and network model combination allow coaches and management to predict how their team is affected by a variety of changes, including trades, substitutions, or alternative plays. Coaches are able to input different line combinations into the model giving them the ability to determine the most effective way to use their players in order to improve their offensive performance. This system also provides the opportunity to determine which plays are the most effective against any given opponent. Teams can determine how players would work together, giving them increased insight when making personnel decisions such as who should be traded. Management can input trade prospects into the model to determine if their abilities would be effective in improving the team as a whole.
This application is just waiting for the tracking information to become available. It can also be adapted to look at defensive capabilities.
How mature is your team’s analytics program? Take the Sports Analytics Maturity Assessment.
Learn about the Groundbreaking Sports Analytics Model coaches and sports analysts are talking about!
Learn all about Sports Analytics here.