Chapter 12 Homework

Read Case Study – Page 357-58: Analytics in the National Basketball Association

Answer Question 1, 2 & 4 only.

Analytics in the National Basketball Association

Historically, the primary method of assessing player or team performance was the “eye test.” This term refers to the intuitive feeling that people with long experience in a sport acquired from watching games and practices. Today, professional sports teams are utilizing analytics to provide data-driven judgments.

The analytics revolution in professional sports started in baseball, as highlighted in the book and the movie Moneyball. Baseball is a fairly easy sport to analyze statistically because it involves a series of one-on-one match-ups between a batter and a pitcher. Furthermore, each play has an obvious start point and end point. In statistical terms, each event is called a “state.”

In contrast to the discrete play in baseball, basketball is one continuous state. Players move instantly from offense to defense. Moreover, regardless of a player’s position, he or she can be at any spot on the court at any time. Analysts could not statistically calculate the odds of (i.e., predict) a particular result occurring (e.g., a player making a shot).

Consequently, modern analytics revolves around the locations and trajectories of the players and the ball. Essentially, analytics in the National Basketball Association (NBA; www.nba.com) is a mapping and data visualization challenge. The challenge is to graphically represent information about movement through space and time; that is, to make data observable.

The First System.

Kirk Goldsberry, a longtime basketball fan with a PhD in geography, undertook the task of developing analytics software for professional basketball. First, he imposed a grid over the usable area of the court where players actually shoot—the area that stretches from just outside the three-point line (roughly 25 feet) to the basket. Then he searched for data in each cell of that 1,284-square-foot grid.

Obtaining the data needed for accurate analysis was difficult. Mapping 10 constantly moving players is not a simple process. On ESPN.com, Goldsberry grabbed statistics for all 700,000 shots taken in every NBA game from 2006 to 2011. The site had such characteristics as the player name, where he took the shot from, and whether it went in the basket. Goldsberry then developed a database mapping the coordinates for all the shots taken in that period.

Goldsberry next analyzed his data to graph shot statistics for individual players, including where a given player shot, how often, and whether or not the shot was good. This system, called CourtVision, revealed player differences that had not been previously detected. For example, Ray Allen, one of the league’s top shooters, had several areas from which he made a high percentage of his shots from the three-point range. However, he rarely tried any mid-range shots.

CourtVision enabled fans, and the league, to visualize their favorite players’ shot patterns. However, CourtVision did not consider factors such as who the defender was or other events going on around the shot. Nevertheless, Goldsberry’s system provided team management with an initial player assessment tool.

Today’s System.

The next opportunity to collect data came when a company called Stats (www.stats.com) developed a six-camera system for basketball. The camera system, which is now employed in all 29 NBA arenas, tracks each player on the court throughout every game. It therefore provides a complete view of the entire game, including tracking individual players and ball possession.

Stats offered its data to Goldsberry. The data were more specific than what Goldsberry had obtained from ESPN.com. Once Goldsberry had the data, he could analyze them to answer any number of questions.

• Players who “draw the defense” can be quantified as ones who pass the ball effectively when two or more players are guarding them.
• “Getting good spacing” visualizes which players control which parts of the court.
• “On-ball defense” assesses how effectively a player defending the ball decreases his opponent’s chance of scoring.

Analyzing the camera data also helped Goldsberry better grasp one of the most difficult parts of analyzing a basketball game: defense. Historically, teams had used basic numbers—for example, how many steals, how many blocks—to evaluate a player’s defensive game. The new system provided a much more refined view of the game.

Goldsberry began by noticing that the space around the basket is the most crucial to defend because this area is where offensive players shoot the highest percentage. Therefore, he analyzed how effectively defenders could stop opponents from making the shot within five feet of the basket. He found that NBA defenders allowed an average shooting percentage of 49.7 in that area.

Utilizing his new data, Goldsberry categorized defenders into two types. The first category blocked or altered their rivals’ shots. In other words, they decreased “shooting efficiency.” In the 2014 NBA season, for example, Indiana Pacers center Roy Hibbert and Milwaukee Bucks center Larry Sanders led the NBA, reduced opponents’ shooting efficiency to 38 percent.

In the second category of defenders, Goldsberry found that they decreased the frequency of their opponents’ shots, in addition to their efficiency. He examined the average rate of shots versus the rate when certain players were defending the area. This way, he could see who was in play when the number of shots decreased. Again in the 2014 NBA season, the best player at this type of defense was Houston Rockets center Dwight Howard, who decreased opponents’ shooting frequency by 9 percent. As opponents shot less often around the basket, they took more mid-range shots, which are the least successful shots in the league.

Because basketball has no states, Goldsberry essentially created them by dividing games into slices of time. He then employed the same types of analyses that had been applied to the states in baseball. He could then enumerate—in terms of points—every player’s every move, from an entry pass into the post (the area close to the basket) to a drive to the basket. These analyses created a new method to assess everything a player and team does.

Let’s consider one example, the Houston Rockets. Utilizing the results of its analytics software, its players almost never attempt long-range two-point jump shots because the Rockets feel this type of shot is among the worst plays. The position is too far from the basket to have a high likelihood of going in, but it’s not far enough (behind the three-point line) to gain an extra point for the risk in taking an even longer shot.

Analytics are impacting not only the Rockets, but the entire NBA as well. For example, in one particular month, NBA data analysis revealed that players attempted more three-point shots than free throws. In fact, the three-point shot defines the shift to analytics in the NBA. It is not a coincidence that the Golden State Warriors, who won the 2015 championship, were the NBA’s top three-point shooting team in the regular season.

Sources: Compiled from “NBA Teams That Have Analytics Department,” NBAStuffer.com, February 18, 2017; T. Ross, “Welcome to Smarter Basketball,” The Atlantic, June 25, 2015; K. Mehta, “Data and the NBA: A Slam Dunk Approach to Basketball,” Umbel, June 22, 2015; C. Benjamin, “The 4 Fallacies of NBA Analytics,” Men’s Journal, June 3, 2015; S. Hammer, “Analytics Key to Modern NBA Success,” The Miscellany News, May 6, 2015; M. Lawrence, “Big Data’s Air-Ball: Five Questions about Players that NBA Analytics Can’t Answer,” Forbes, February 27, 2015; B. Alamar, “The Inside Man: NBA Analytics,” ESPN.com, February 19, 2015; M. McClusky, “One Man’s Quest to Track Every NBA Remade Basketball,” Wired, October 28, 2014; T. Moynihan, “The NBA’s New High-Tech Control Center Is a Hoops Fan’s Dream,” Wired, October 28, 2014; R. Simmons, “Golden State Warriors at the Forefront of NBA Data Analysis,” SFGate, September 14, 2014; B. Holmes, “New Age of NBA Analytics: Advantage or Overload?” Boston Globe, March 30, 2014; D. Oliver, “How Numbers Have Changed the NBA,” ESPN.com, November 15, 2013; www.nba.com, accessed August 25, 2016.

Questions

1. Provide an example of the use of descriptive analytics for an NBA team using Goldberry’s system.
2. Provide an example of the use of predictive analytics for an NBA team using Goldberry’s system.
3. Provide an example of the use of prescriptive analytics for an NBA team using Goldberry’s system.
4. What are the advantages and disadvantages of Goldsberry’s system to NBA players? Provide specific examples to support your answer.