‘Hero ball’ describes the claim by some fans that a team’s star players hog the ball more during playoffs than regular season. (Steve Lipofski/Basketballphoto.com)

Do NBA teams play differently during playoffs? Hardcore fans insist that star players are more likely to hog the ball and take long, risky shots. From informal observation, fans have come to name this very un-teamlike behavior ‘hero ball.’

It now turns out hero ball could be a real phenomenon. Crunching two seasons of NBA data, a team of students found that playoff teams passed the ball less, dribbled more, and held the ball longer than during regular season. Their analysis won first place at an NBA-sponsored hackathon last month.

“If a team is going down in the playoffs, they’re going down with their best players taking the shots,” said team member Alex Wainger who is studying for a master’s in data science at Columbia. “LeBron James and Kyrie Irving with the Cleveland Cavaliers are notorious for this.”

A native New Yorker and longtime Knicks fan, Wainger applied to compete in the NBA Basketball Analytics Hackathon the instant he saw a posting for the event. During orientation, he talked another first-year data science student and basketball fan, Zac Robertson, into joining him.

They got to review the datasets a few days before the Sept. 24 hackathon, but hit upon their winning idea a few hours in, on game day. “Hero ball in the playoffs had been a huge topic of conversation for several years,” said Wainger. “We liked the idea of trying to verify or refute conventional wisdom.”

With more than five gigabytes of NBA data, the winning team showed that ‘hero ball’ is a real phenonmenon. In the visualization above, they found that average touch-time before a shot was greatest during playoff season.

Joined by two others, Ella Kuzmenko and William Robbins, undergraduates at NYU and Pace University respectively, they sorted through more than five gigabytes of data, including statistics on shots and possessions, and measurements of exact player locations on the court.   

They started by translating hero ball into numerical terms. From dribbles-per-possession to touch-time-before-a-shot, they found evidence suggesting more selfish play during playoffs. They then built an interactive graphic allowing anyone to explore how these values change from regular season play to the playoffs.

They also found that teams scored fewer points during the playoffs. Defined by less teamwork, hero ball was probably one factor, they concluded. But other explanations could include greater familiarity with the opposing team’s offense from prior games, and the greater intensity and preparation that go into such high-stakes games, they say.

A team of students from Columbia, Pace and NYU took first place at the NBA’s first hackathon on Sept. 24. (NBA)

They presented their findings that evening in a slide presentation titled “The Case Against Hero Ball.” Competing against four other finalists, they came in first place. The top prize includes tickets to an NBA game, a $200 gift card to an NBA store, and an open invitation to play pickup basketball at Terminal 23, the hackathon venue. But the one they are most excited about is their lunch with NBA Commissioner Adam Siler and other executives next month.

The data revolution has been slower to reach basketball than other sports because of its fast pace and relatively fluid play, but the NBA is working to change that. It recently invested in cameras for all of its arenas to capture subtle aspects of play that previously went unmeasured. Its first hackathon received favorable reviews from TechCrunch and the basketball data analytics blog Nylon Calculus.

Read interviews with the winning team in TechCrunch:
The NBA’s first hackathon: How is the future coming for the league?

— Kim Martineau