DU Hosts the Rocky Mountain Sports Analytics Symposium

NFL Director of Data and Analytics Mike Lopez

Ryan McShane flew in from Dallas, Karl Pazdernik came in from Washington state and Jacque Kvam travelled from the Bay Area. They all wanted to hear about the latest research in sports analytics, so it was a priority for them to attend the Rocky Mountain Sports Analytics Symposium (RMSAS) hosted by the Department of Business Information and Analytics at the Daniels College of Business Aug. 2.

The day was packed with speakers who shared with the audience of more than 120 their expertise and latest research on several sports, including football, basketball, hockey and baseball.

Pazdernik, co-founder of Deep Football, presented his paper, “A Deeper Understanding of Quarterback Pressure in Football.” With a PhD in statistics, Pazdernik noted that he could research any topic, but he has a passion for football.

“It’s an exciting time with the new player tracking data out,” he said. “It’s fun to play with it.”

The keynote speaker at the RMSAS spoke extensively about that new tracking data. After all, he is the director of data and analytics for the NFL.

Mike Lopez explained that beginning in 2016, the NFL placed radio-frequency identification chips in the shoulder pads of each player, and on the ball. So, data analysts for the teams can now track the real-time location, speed and direction of the ball and each player.

While access to this new data is wonderful, Lopez stressed its limitations.

“Sports data is observational,” he said. “You can’t randomize [data] in football.”

Lopez elaborated by citing a medical study on hormone replacement therapy (HRT) for women with breast cancer. In an initial study, researchers showed that there was a higher risk of death in women who took HRT. But, when a separate set of researchers conducted a randomized trial, there was no rise in death risk to women who took HRT. The risk was due to something unknown that the initial researchers didn’t measure.

Applying this notion to the football world, Lopez mentioned a famous case by economist David Romer, whose 2005 paper proved that more coaches should go for it on fourth down.

Lopez showed the audience snapshots from two 2013 games where teams faced different fourth-and-1 scenarios. He had the audience guess what was different about the two NFL games that seemed like identical decisions for the coaches. Hands flew up. Guess after guess. No one was right.

Lopez let the audience in on the secret. The difference was something that couldn’t easily be seen on the screen. While both coaches were facing what appeared to be the same fourth-and-1 scenario, the data showed that one team was really facing fourth and 0.21 yards, while the other was facing fourth and 1.58 yards.

Assistant Professor Ryan Elmore

“You think both coaches are calling [a] fourth-and-1 [play], but they’re not. We [researchers] thought the distances were the same, but they weren’t,” Lopez said. “The coaches are looking at the actual distance.”

His point is that researchers might have been unfairly critical of coaches who didn’t go for it on fourth-and-1 situations. But now with RFID, the researchers will know that not all fourth-and-1 plays are equal. And yes, even with the tracking data, Lopez still has reservations.

“Player tracking is not everything. No player is the same. No play’s the same. We will never be able to account for everything,” he said.

While Lopez was the keynote, speakers shared observations throughout the day from a variety of sports.

“The RMSAS was a wonderful success for the University and the individual researchers alike,” said Ryan Elmore, assistant professor of business information and analytics and organizer of the RMSAS. “We brought some of the world’s leading analytical experts in football, basketball, hockey, tennis and e-sports, among others, together to discuss the latest state-of-the-art analytics in their respective domains.”

McShane, a PhD candidate in statistical science at Southern Methodist University, is doing his dissertation on modeling on head-to-head competition outcomes. He doesn’t plan on immediately going into a sports analytics job, but will teach statistics at Amherst College in the coming semester.

“The conference was a good opportunity to meet, or catch up with, others working in the area,” he said. “I enjoyed the interview with the NBA analysts especially. It was informative and interesting to hear the analysts talk about their work.”