Basketball and Data background
The Houston Rockets made an interesting discovery several years ago: data can transform when and where his players decided to take shots, and what type of players to recruit for the team. Historically, however, data was gathered based on reported player statistics and performance. Now data is being gathered in real time from visual imagery on the field of play.
A Bit of Background
In the late 1990s, the Oakland Athletics General Manager, Sandy Alderson, had to do something as the owners decided to slash their payroll. To field a competitive roster on a limited budget, Alderson began focusing on a form of data analysis they called “sabermetric” to obtain undervalued players (“sabermetrics” is the empirical analysis of baseball, especially baseball statistics that measure in-game activity.). The metric he found to be the most successful: on-base percentage among hitters. Later, the famed general manager Billy Beane would be profiled in Michael Lewis’s 2003 best-selling book “Moneyball: The Art of Winning an Unfair Game.”
Over 15 years ago, Lewis discussed Beane’s methods as the General Manager of the Oakland Athletics and how Beane — along with Harvard-educated statistician Paul DePodesta — used sabermetric principles to run his team in a cost-effective way. The book and Beane’s methods have influenced the way many think about the game of baseball, and as importantly to sports data in general.
Basketball and Data
Data is being used the change the way everything is looked at, including basketball. Now, unlike in the late 1990s, in 2020 artificial intelligence and machine learning tools for analyzing and visualizing data has taken a massive step ahead. Companies of all sizes and varieties are getting into the sports data game.
In this video, Darryl Morey, the General Manager of the Houston Rockets basketball team, makes it clear: “Analytics has permeated everything.” By using a video tracking system to analyze player move, teams are able to identify which shots have the greatest value. Morey’s conclusion was that these are the two most valuable shots, and ensuring you have players make those is statistically the best way to win:
- 2-point dunks, and
- 3-pointer shots (vs. 2-point shots from within the line)
Second Spectrum, one such sports data analytics company, discusses how they gather and analyze data for the NBA teams through cameras that now track and record 3 dimensional spatial data for every player and every ball movement. Machine learning is used to create interactive visualizations to help understand what works and what doesn’t in basketball for the players. As the company’s owner say: The machine” has become the integrated, best-practice store of what helps elite players and teams to win. The machine puts an analytical data point on what previously was a feeling about the success of a specific shot or move.
One of the key elements of developing AI algorithms is learning (or deciding) what matters for each use case. Historically, even when this video was made, the user needed to decide what mattered. Now, intelligent systems learn from data without upfront human intervention. Algorithms become evolving tools to understand what matters and why.