When it comes to artificial intelligence, the applications are endless. However, at the intersection of AI and sports, there are clear ways in which machine learning and other forms of AI are able to revolutionize the way a game is played.
There are a variety of ways in which to apply machine learning to a sports league. There are programs created to train athletes, such as VR and AR platforms that allow players to practice specific situations.
These programs are similar to familiar video game franchises like NBA2K20, which are associated with a specific league. However, unlike a video game RPG, machine learning (ML) platforms are specifically tailored to develop a certain skill, like a free throw, by emulating exact conditions.
There are other programs that track players and report stats to officials. These types of programs help coaches and other staff make selections when it comes to a draft or recruitment scenario.
And lastly, there are ML applications designed to alter how data is collected, analyzed, and applied. While the NHL, MLB, PGA, and NFL are also involved in the race to unite AI and sports analysis, the NBA has the most direct applications in play at the moment.
Converting Data Points
The advent of sports technology has resulted in a vast amount of hard data for statisticians and analysts to sink their teeth into. However, there’s yet to be a concrete application for the array of data points collected.
In fact, it may be punters and sportsbooks who are most familiar with cross-analyzing LeBron James’ basket percentage to Giannis Antetokounmpo’s block percentage. For those who follow NBA betting odds all season long, these data points are a basic part of the famous point spread. For scientists, these stats are working to integrate this data into a machine learning platform.
Once this data is collected and aggregated, AI and ML programs draw conclusions from the different fields of information. Sets of data are created from key basketball players, the court itself, or a team at large in order to note patterns and create an analysis.
In each game during the NBA’s regular season, around 12,300 data points are collected. Every regular season runs for 82 games, which means that each year there are over 1 million data points collected from the NBA alone.
Computer Visualization (CV) in the NBA
Computer visualization is one way that millions upon millions of data points can be applied. Certain programs are able to analyze images and videos in order to replicate a human system. In this case, the human system is the NBA’s version of basketball.
Motion-tracking cameras are able to collect data to create profiles on each player, team, and court. As aforementioned, this type of computer visualization technology can be used by staff to make quick assessments of players and matches.
In fact, these profiles are becoming so standardized and reliable that they’ll likely replace the in-depth analysis now created by sports analysts. However, it’s likely that a human touch will still be required to overview each report made by CV technology.
Time Series (TS) in the NBA
Time Series technology studies specific sets of data in relation to a specific segment of time. This is used to predict outcomes based on in-depth patterns and other characteristics of long-term data points.
TS technology will be specifically relevant to the punters we mentioned earlier. While anyone betting on the NBA knows which stats are most relevant when creating a handicap between teams, TS will be able to use deeper historical data.
Currently, few sportsbooks analyze play from the 1970s Celtics because that specific team is no longer playing and, therefore, irrelevant to modern predictions. However, TS technology would be able to cross-reference data points and patterns across each player, coach, team, division, and the outcome of each intersection across decades.
In short, TS technology could predict the peak and fall of dynastic teams, individual players, and divisions at large. However, this type of retroactive analysis comes with a minefield of red tape.
Specifically, the NBA will need to navigate issues with collecting information and consent to generate a specific report.