Exploring the Analytics Behind Better Sports Decisions
Eric Bradlow has all the grit and grind of your toughest football linebacker. His equipment of choice? Data and analytics.
As an applied statistician, Dr. Bradlow, a professor of marketing, statistics, education and economics and vice dean of analytics at the Wharton School of the University of Pennsylvania, uses high-powered statistical models to solve problems. Combining math and observed data, these models use statistical assumptions to generate sample data, find patterns and make predictions that help businesspeople make data-driven decisions on everything from the best products to sell, to the optimal marketing messages.
Sports Analytics in Motion
In the business of sports, statistical models have real muscle. Deep data analysis helps with risk assessment and strategy formation on and off the playing field. Dr. Bradlow, who has worked with the Philadelphia Eagles football team, has observed the power of data analytics in player evaluation, on-field calls, injury-prevention, training and, he says, every aspect of sports. It is a tool that all teams, leagues, coaches and scouts are adopting, if not embracing.
“At parties, people ask what I did for the Eagles, and I describe it like this,” said Dr. Bradlow during a conversation on sports analytics with Wharton Dean Erika James. “When the Eagles have five minutes on the clock to make a draft pick and a scout uses their judgment and says, Mr. Lurie [CEO of the Eagles], Mr. Roseman [general manager of the Eagles], we should pick this player. I want [Lurie and Roseman] to be able to look on their screen and say this is what Bradlow’s algorithm says the success of that player will be. And then to go back to that [scout] and say, ‘I want you to know that the data says something different. You have 30 seconds to tell me why the data is saying one thing and your professional judgment is saying something else.’ That is what good managers do. We can’t ignore data. We can’t ignore algorithms. We can’t ignore the predictions. But they are data and information and support tools to help the experts make decisions.”
High school students who attend Wharton Global Youth’s Moneyball Academy and Moneyball Academy: Training Camp are intrigued by this idea of using data to make decisions on the playing field. Sponsored by the Wharton Sports Analytics and Business Initiative (WSABI) and led by Wharton’s Adi Wyner, these summer programs prepare students to be leaders in an increasingly data-driven economy.
With WSABI’s help, Wharton Global Youth asked past Wharton Moneyballers to share their ideas about the power of sports analytics, including insights from their projects that were published in various editions of the Wharton Sports Analytics Student Research Journal.
A common thread: their love of sports, their reverence for data, and their genuine enthusiasm for putting both into practice.
Stolen Bases and Running Backs
Nate Y. has always been interested in what he calls “the crossroads of math and sports” – like, always. As a kindergartner (so his parents tell him), he would wake up, run to his computer, and check out all his favorite teams’ box scores and game logs on ESPN.
Fast forward about 13 years and Nate, now a high school senior at The Leffell School near Scarsdale, N.Y., has gone deep on statistics in football and baseball, his two favorite sports, and has improved his analysis after learning skills like R programming language for coding during Moneyball Training Camp and Moneyball Academy. “Learning statistics literally transformed my high school experience,” notes Nate, who was accepted early decision to Wharton, where he plans to attend this fall. “My high school doesn’t have a stats class. After my 9th grade summer in Moneyball, I was so fascinated that I had to do more. I self-studied AP Stats as an independent course and then in 10th grade started a football analytics blog with my friend. My Moneyball experience also served as a springboard to further study R and apply it to advanced statistical projects such as creating a lexicon for analyzing emojis in financial discourse on X and creating SIDELINED, an R-Shiny interactive app that empowers others to challenge gender biases in ESPN NCAA basketball journalism.”
While in Moneyball Training Camp, Nate and his three teammates worked on the project Stolen Bases Are Disappearing as Baseball Becomes More Analytical. Using the Lahman dataset of Major League Baseball pitching, hitting and fielding stats, they created a scatter plot between successful steal rates and runs scored. “We found that most times it’s not worth it for players to steal bases,” says Nate.
Nate is more excited to discuss the Moneyball project he tackled the following summer when he and a few classmates rallied around one of their favorite pastimes: fantasy football. “I’m a huge fantasy football player and manage five teams,” says Nate, referring to the game in which participants serve as owners and managers of virtual American football teams. “We wanted to figure out predicting running backs, which is an important position in fantasy football…We ran a multivariate regression model, made our rankings and then we figured out which players our rankings favored compared to Yahoo! and ESPN.” The model helped Nate draft two winning running backs for his fantasy football league.
Nate’s best advice for aspiring analysts: “Find something that generally makes you extremely excited and then bring different disciplines that interest you together. Sports with math ended up being transformative for me.”
College Football Recruiting
A native of Los Angeles, Calif., Naya K. grew up a sports fan who loved her hometown teams. But when she moved to Fayetteville, Arkansas, she was faced with a challenge: Arkansas has no professional sports teams. “I became a huge fan of college athletics, especially college football,” says Naya, a high school senior at Haas Hall Academy.
So, when Naya ended up at Wharton’s Moneyball Academy the summer before her junior year, she quickly teamed up with her fellow college-football fans to tackle a sports-analytics research project. During brainstorming sessions, they noticed a void in college football rankings: no system existed to evaluate programs on player development. They set out to create a statistical model that ranked programs on their ability to develop players, taking into account high school recruiting rankings and American National Football League (NFL) Draft results.
The group detailed its analytical approach in the paper Stars Matter: An Analysis of College Football Recruiting, Development and Draft Success. “The project was one of the most exhilarating things I’ve been a part of,” recalls Naya, who is working with the University of Arkansas softball team as a data analytics intern. “We pulled data from 24/7 Sports and , Draft results to create a regression model. The regression model showed the predicted NFL Draft result given a player’s high school recruiting ranking. The difference between the predicted and actual draft results we attributed to the player’s development in college. By aggregating the residuals of each college program, we were able to determine which programs were the best at developing their players for the NFL Draft. Using the data, we concluded that The Ohio State was a program that was exemplary at developing their players.”
Naya says she has always been someone who believes in the truth in numbers and appreciates logic and objectivity when making decisions. “In sports, there are so many layers of decisions from recruiting to game strategy to player development,” she observes. “Data analysis is a great way to examine a variety of factors when making these decisions and gain a perspective beyond the traditional way we evaluate sports.”
Predicting the Quarterback MVP
Ryan W.’s dream job is to be a general manager in the NFL. It’s safe to say that he is making serious strides toward that goal, led by his mantra: passion is the single greatest driver of achievement. And it all started with a stint a few years back in Wharton’s Moneyball. “I didn’t know I had an interest in statistics or sports analytics, but once I got there, I absolutely loved it,” says Ryan, a student at Indiana University’s Kelly School of Business.
During Moneyball Academy, Ryan, then a high school student at Eastside Preparatory School in Seattle, Washington, teamed up with Penn PhD student Ryan Brill to further explore their interest in sports analytics through research. The result: Predicting the Quarterback MVP in the NFL, an analytics project in which the Ryans created a logistic regression model to predict the NFL Quarterback MVP or Most Valuable Player. “The NFL MVP award is chosen each year by a panel of 50 sportswriters who are selected by the Associated Press. As the MVP is chosen by humans who do not necessarily base their decision on statistics, but on watching all the games and talking to coaches and players, it is natural to wonder whether there is a mathematical rule that can describe the MVP selection process.”
Their algorithm predicted the correct quarterback MVP in each year starting in 2003, except 2009 and 2015, when their results pointed to two players whom the popular media agreed were snubbed. It was Ryan’s first opportunity to really go deep on the data. “I had a solid coding background and was able to understand the mathematics of statistics much further working with Ryan Brill,” notes Ryan. More importantly, he connected with Moneyball lecturer Eric Eager, who was then with Pro Football Focus (PFF), a sports analytics company that analyzes the NFL. Ryan got to know Eric so well that he interned at PFF for two years during high school.
All of these data-driven experiences helped to lay the foundation for Ryan’s college career as a finance and business analytics major, which has also included doing analytics for the Indiana football team. “In football specifically, data and film are the two pillars that hold up a football scouting or information program. You have to couple anecdotal evidence with data,” he says. “So, when you see something on the field in football, like when [the other team] runs to the right and a certain linebacker performs much worse at filling that gap, but you’re not sure that is true across all things, you use data to answer that question. Similarly, if you notice a quarterback performs worse against a certain coverage, you watch film to see where exactly he fails. Data is incredibly powerful because it’s concrete information. If done right, it empowers you to make so many more successful decisions.”
Conversation Starters
Describe Dr. Eric Bradlow’s job as an applied statistician. What does he do and who has he worked for?
Nate’s best advice is to “Find something that generally makes you extremely excited and then bring different disciplines that interest you together.” How might this advice fit into your own interests? How could you make your future pursuits more multi-disciplinary?
Which Moneyball sports-analytics project intrigues you the most? What questions do you have about areas of sports that interest you? How might data and analytics provide clarity?