Data Analytics in Basketball

Davidson’s Peyton Aldridge hustling against St. Louis (courtesy of Davidson Basketball)

In basketball, you often hear of players that have the “intangibles” or a “knack for the ball.” Some might not fill up the box score with points, rebounds, and assists, but still get consistent playing time because they help the team.

Shane Battier, Duke Blue Devil legend, is a prime example. He did not always have the stats to back his minutes on the court. However, he played in the NBA for 13 years and won two championships. Why? When he was on the court, his team played better.

These qualities are often lumped into the catch-all “hustle.” Battier’s hustle reduced the scoring opportunities of the other team and created plays for his teammates.

So, if hustle is the product of certain intangibles without a specific formula to compute, how can we quantify such an integral part of a team’s success? This question fascinated me and spurred me to look further into it last summer.

The NBA recently started tracking certain hustle statistics:

  • screen assists (setting a screen that leads to a made basket)
  • deflections
  • loose balls recovered
  • charges drawn
  • contested two- and three-point shots

This was a great start to help quantify hustle, but there were two problems: I believed that other statistics could help quantify a player’s hustle, and the NBA didn’t consolidate these statistics into one score. Therefore, I couldn’t holistically compare one player’s hustle to another.

To tackle the first problem, I assembled a list of additional statistics that I thought contributed to hustle. After consulting Dr. Tim Chartier (Davidson Math and Computer Science Professor), Dr. Mark Foley (Economics Professor), and the men’s basketball coaching staff, we came up with the following:

  • steals
  • blocks
  • offensive rebounds
  • contested defensive rebounds
  • defensive field goal differential (the difference in an opponent’s shooting percentage when guarded by a certain player)

Now, with all these different statistics, I wanted to combine them to create a single hustle score per player. Fortunately, a 2015 Grantland article gave me a head start. Author Jason Concepcion assigned “hustle points” to each of the five statistics the NBA already tracks. These points are based on the probability of the action occurring multiplied by its expected point value. For example, if a two-point shot is contested, the probability of the shot going in decreases by 0.05. Thus, the hustle point value is 0.05 x 2, which equals 0.1 hustle points. I adapted this method into my model. Further, summing up all the hustle points from each statistic for a player created a single, cumulative “hustle score,” so I could compare different players.

According to the model, these are the 10 best hustlers from last season:

10 Best Hustlers from 2016-17 NBA Season

Player Hustle Score
Giannis Antetokounmpo10.32
Draymond Green8.48
Anthony Davis8.48
DeMarcus Cousins8.46
Patrick Beverley8.12
Marcus Smart7.83
John Wall6.06
Jrue Holiday5.53
Kemba Walker5.98
Robert Covington5.44

Because of the formula I adopted from the article, these cumulative scores represent the amount of points each player created and saved per game through hustle.

Boston Celtic Marcus Smart hustling against Philadelphia (courtesy of ESPN)

Some of these players, like Anthony Davis and Draymond Green, are superstars known for their defense and energy. Others, however, are much less well-known. Marcus Smart is a fascinating example. He doesn’t come close to topping the league in any of the common statistics, like points, rebounds, or assists, but he plays some of the most minutes on the Boston Celtics. During one game, the announcers were confused about why a player who tends to shoot well below average plays so much, and whenever he’s on the court, his team plays better. Well, now I found an answer: his hustle.

The ability to quantify hustle with a single statistic should revolutionize the basketball world. Like with Marcus, it validates a player’s playing time to coaches, fans, and players. This statistic also adds another dimension to scouting. Recruiting, drafting, and trading players all require analysis of information to make important decisions. As shown above, players with high hustle scores have a significant impact on team performance. Therefore, the addition of the hustle statistic should play an integral role in team-building.

Leave a Reply

Your email address will not be published. Required fields are marked *