The NBA Bets Big on Data

Arup Sen (GRS’13) analyzes gambling and basketball data to ensure the game’s integrity

| in Alumni, Features

By Marc Chalufour

As a kid in Illinois, Arup Sen eagerly awaited the arrival of the evening newspaper. His interests in sports and numbers came together in the agate type of the box scores, where he could see how the Los Angeles Lakers or the Oakland Athletics or any other dominant team of the 1980s had performed the night before. If his parents felt they had to punish him, they didn’t send him to his room—they took away the sports page.

Sen (GRS’13) followed basketball and baseball in the US and, when his family moved to India, he adopted soccer, cricket, tennis, and badminton. He couldn’t imagine a career in sports then—but by the time he was completing his economics PhD at BU, a statistical revolution had swept through professional leagues. Analysts, data scientists, and economists were in high demand.

In 2014, Sen landed a dream job at the National Basketball Association (NBA), analyzing referee performance. His rise through the organization since then reflects the growth in scope and sophistication of the league’s analytics operation. Today, Sen is the senior director of basketball strategy and integrity. With the rapid spread of legalized sports betting, their analysis of player and referee performance in conjunction with trends in gambling markets, plays a critical role in ensuring the league’s competitive integrity.

Sports Betting Goes Mainstream

If you’ve watched a live sporting event recently—and maybe even if you haven’t—you’ve seen sports betting ads. Online sportsbooks like DraftKings and FanDuel jockey with physical casinos for gamblers’ money. Paid experts and analysts offer up their predictions and parlays of the day. This was unimaginable just a few years ago.

Prior to 2018, sports betting in the US was only legal in Nevada, where $4.8 billion was bet in 2017. That same year, an estimated $150 billion was bet on sports through offshore gambling services. The US Supreme Court’s 2018 decision in Murphy vs. NCAA struck down the Professional and Amateur Sports Protection Act of 1992 and opened the door for more states to legalize sports betting. More than 30 states have since legalized some form of sports betting.

 Widespread legal gambling has huge implications for sports leagues—but so did illegal gambling. Members of the 1919 Chicago White Sox—aka the Black Sox—were famously accused of accepting bribes and intentionally losing the World Series; players from seven college basketball teams were caught in a scheme to shave points in 1951, meaning they attempted to prevent their team from covering the point spread; and Pete Rose, baseball’s all-time hits leader, was banned from the sport because he bet on games he participated in. And those are just the most notorious of the scandals. Proponents of sports betting legalization argue that it’s better to bring gambling into the open where the action can be analyzed.

The NBA has been a leader in this transition. Four years before Murphy vs. NCAA reached the Supreme Court, NBA commissioner Adam Silver called on Congress to allow states to legalize sports betting, “subject to strict regulatory requirements and technological safeguards.” Behind the scenes, the NBA was already developing those safeguards.

Leaving the Family Business

Sen’s father taught economics at the University of Illinois Urbana-Champaign and, later, at the University of Delhi. Sen assumed he’d follow him into “the family business.” Not until he was deep into his PhD studies did Sen notice a trend: “All of my research was in basketball,” he says.

His thesis focused on the NBA labor market—how players respond to contract incentives and whether they should sign with good or bad teams. He presented a portion of that work at the 2011 MIT Sloan Sports Analytics Conference, where his analysis of how NBA players perform in the final year of a contract versus early in a long-term deal was awarded best research paper.

That conference helped him build connections in the sports world: ESPN invited Sen to write a column about how to fix the NBA Draft; he spent a few months analyzing contract arbitration outcomes for Major League Baseball; and he began talking with the NBA about analyzing referee performance. But without the prospect of a full-time job, he accepted a position at consulting giant Deloitte. Then the NBA called him back. “I told the person who had hired me at Deloitte that I hadn’t been searching for anything—but this could be a dream job,” he says.

A New Frontier of Analysis

Sen’s first role at the NBA was manager of referee strategy. He would use analytics to develop ways to measure and enhance referee performance and consistency. Since then, his roles have included senior data scientist and integrity analytics lead, and the information he works with has expanded to include player performance and betting data.

Sports have undergone a data-driven revolution in the 21st century. Front office personnel with academic as well as athletic backgrounds have developed increasingly sophisticated ways to evaluate player performance. In basketball, traditional stats like points, rebounds, and assists are still used, but they’re no longer considered the best way to measure success. Advanced metrics attempt to more accurately capture a player’s performance. PER (player efficiency rating), for example, tracks per-minute productivity—and that’s just the tip of the statistical iceberg.

 In 2014, the NBA began using video technology to collect a massive amount of data on every game. Second Spectrum cameras in NBA arenas record the movements of everything on the court—10 players, 3 referees, and the ball. The system generates millions of data points, allowing analysts to study where players and referees position themselves, how fast they do so, and much more. “That data is very granular,” Sen says. “We’re still trying to figure out how to use it in our integrity analyses.”

On the gambling side, the NBA has partnered with multiple sportsbooks. They receive broad data on sports betting trends and anonymized data on individual bets, which they use for integrity analysis only. Sports betting is expanding so quickly that Sen says one of their challenges is to learn the nuances of each market. How do online sportsbooks differ from Vegas casinos? Are there state-by-state differences in betting patterns and behavior? Each distinction creates another layer for analysis.

 Although data science is a major component of his work, Sen still considers himself an economist. “I think of economists as having some underlying thought about what you would expect behavior to be, and then you try and test it,” he says. “Data science is much more empirical in nature: you’re looking for patterns and you come to your conclusions based on what you see in the data, then you come up with a story that’s consistent with that.” Both approaches serve his team well. They monitor growing streams of data for patterns—or disruptions in patterns—but they also consider potential scenarios to watch out for.

Not unlike regulators on Wall Street, the NBA’s basketball strategy and integrity team monitors all this information for signs of overt manipulation and looks for the athletic equivalent of insider trading. In a purely hypothetical situation, if Nikola Jokić—the NBA’s defending MVP—were to get injured, his absence from Denver Nuggets games would impact the point spread and shift betting patterns. In situations like that, Sen wants to know if betting trends change before the injury is announced publicly—which would suggest someone either intentionally or inadvertently leaked valuable information. Or, consider a slightly different scenario: the betting line on Nuggets games begins moving more than anticipated and there’s no corresponding announcement of an injury. That might also raise an alarm. “There should be a basketball reason that explains the line move,” Sen says.

Although Sen won’t cite specific examples of his team’s findings, he says it’s rare that an urgent finding has to be brought to the immediate attention of the NBA’s senior leadership or legal department. And even when they do flag something, closer examination typically rules out any intentional manipulation of games.

Sen notes that early sports betting scandals could be comically simplistic. “It was like, ‘Hey, I’ll fall down six times and hope nobody notices,’” he jokes. The expectation today is that any attempt to alter a game’s outcome, in front of dozens of high-def cameras and thousands of cell phones—not to mention the scrutiny of data scientists—would need to be far more sophisticated.

The challenge for Sen and his team is to keep up with the growing range of data available to them, from new betting markets to complex video analysis to the flow of NBA information over social media. The one constant, he says, is “we’re monitoring way more than you think we are.”