Just before they took the stage, the panel — which included former NFL player and broadcaster Greg Olsen, the NFL’s head of analytics and Amazon Web Services executives — reminded them that their ideas would not stay in the room. Research from the Big Data Bowl directly influences real NFL decisions, shaping coaching strategies, player development and even broadcast coverage.
“I was nervous to make sure we got everything right, but also really excited,” Nielson said. “This is something that’s going to move the game we love forward. Millions of people care about the NFL and this matters to the NFL as a whole. And I realized, ‘Me and my friends are a critical part of that, at least for the next hour.'”
Just as a defensive back has only fractions of a second to decide whether to break on a route or stay home, Miller, Nielson and Thompson knew they had one opportunity to show how those fractions of a second can now be measured.
“We answer the question of what a cornerback or a defensive back could have done and what would have happened if he had done things differently,” Thompson said. “If you think he should have jumped the route, our computations show where he could have gone and the probability that he could have gotten an interception. Data. Not just vibes.”
For fans, that play happens too fast to process. For coaches, it can be the difference between winning and losing. The team’s model now translates tracking data points to provide the league with something official: the realistic options a defender actually has once the ball has left the quarterback’s hand.
Building the model was tough, but turning the complex statistical modeling into something fans and NFL front office executives could understand was a monumental task.
“We’re statisticians. We can do the statistics, but a video? We wouldn’t even know how to approach something like that,” Miller said.
That’s why this year the team recruited a different kind of player — freelance motion designer Evan West. According to Thompson, the project would have been impossible without West’s feedback and expertise.
BYU master’s students have competed in the Big Data Bowl Competition before, including some members of this year’s winning team.
“Last year we worked really hard on the project,” Nielson said. “But a couple of things in and out of our control didn’t go right and we didn’t win. The whole idea of coming back this year was, ‘Hey, let’s give it our best shot again.'”
While the experience was one in a million, the team believes anyone could do it next.
“It’s not like our work wasn’t well done, but you don’t need to be a crazy genius to do what we did,” Thompson said. “You just have to be willing to put yourself out there, put in the time and be willing to fail. We competed last year. Didn’t work out, but we learned a lot from it, and we used that momentum to improve. We’re still just BYU students. We might be in a master’s program, but we were all undergrads and learned everything here.”
“We’re opening doors for other people,” Nielson said. “There’s some really great sports analytics research going on by professors, graduate students and undergrads here at BYU. And we’re showing them, ‘Hey, we don’t have the dedicated sports analytics program some schools have — we’re learning the fundamentals, we’re learning the theory, but we can also hang with the big guys.'”
