Analytics Drives Next Generation Of Moneyball In Sports
Professional sports teams now use deep analytics to drive decisions about nearly every aspect of the game. A recent conference drew practitioners -- and job seekers in this new career path.
![](https://eu-images.contentstack.com/v3/assets/blt69509c9116440be8/bltc0182b2356ae8eed/64b83949410a1b4c0bd7459b/IW_generic_image.png?width=850&auto=webp&quality=95&format=jpg&disable=upscale)
The Next Big Stats
MIT's prestigious Sloan School of Management hosts the Sports Analytics Conference in part because sports teams have become such high-profile adopters of a numbers-driven management. There are valuable lessons--not to mention a potential career path--for up-and-coming business majors, and some 700 data-minded students were among the attendees.
In a research track tied to the event, PhD students, recent grads, and even a few professors presented papers on the next big analytics that might be used by professional sports to make better business moves. In one of 15 finalist papers, Robert Ayer, a 2011 grad of MIT, presented a paper looking at what combination of basketball talent -- a big man, a point guard, a swingman -- delivers the most wins.
Analyzing more than 30 years of NBA game data, Ayer segmented players by type and found that certain combinations of top-two and top-three players on NBA teams have consistently delivered more wins, regardless of the influence of individual player talent or coaching (factors that were statistically cancelled out in the analysis).
[ Want more on analytics? Visit our 12 Top Big Data Analytics Players image gallery. ]
Ayer found that the best two-player combination is a versatile, 3-point-shooting wing and a high-scoring, high-rebounding center. That delivered 7.59 incremental wins per season. Conversely, having two high-scoring wings reduced wins by 4.05 games per season.
Looking at the top three players, the combination of a high-scoring, high-assist point guard, a versatile, 3-point-shooting wing, and a high-scoring, high-rebounding center was found to yield 13.6 incremental wins per season. Conversely, the combination of two dynamic power forwards and a low-scoring, pass-first point guard diminished wins by 8.47 games per season.
This year's winning paper, "Deconstructing the Rebound with Optical Tracking Data," by Rajiv Mahsewaran, Yu-Han Chang, Aaron Henenhan, and Samantha Danesis, all of the University of Southern California, made use of optical tracking data provided by Stats Inc.'s SportVU technology to show how shot location and the positioning of offensive and defensive players affect rebounding. Among the conclusions: three-point shots and close shots are much more productive sources of rebounds than mid-range shots (due to factors including the trajectory of the ball and typical positioning of offensive players in these scenarios), and potential rebounds are clustered much closer to the basket than experts predicted.
Stats Inc.'s SportVU is the NBA equivalent of the Pitch FX system used by baseball since 2006 (and soon to joined by a Field FX system that collects data on fielding). These camera-based systems track the position of players, the ball, vectors, and velocities on every pitch and play, generating huge amounts of data in the process.
"It adds up to more than a million data points per NBA game," said Kevin Goodfellow, an executive of Sportsdatahub.com. Presenting on "The Revolution in Advanced Sports Analytic Systems," Goodfellow predicted that the same open source technologies that businesses are using to make sense of their big data hordes, like Hadoop and NoSQL databases, would become staples of sports analytics as optical data systems and other data-intensive sources become commonplace.
As usual when it comes to analytics, even the fire hose of data baseball and basketball execs now have only makes them thirsty for more. Bill James of the Red Sox wants more and better data on prospective players. "Most of the information that we have is about Major League players, but most of the information we need is about players in college, players in Japan, and players at the Double A level," James said. "If we say that the Major Leagues are at level 1, is Triple A at .8 and a good college program at .6? We really don't have a clue."
Despite the huge strides made in sports analytics over the last decade, the emergence of more data and new analyses in the decade to come will undoubtedly make today's insights look primitive.
About the Author(s)
You May Also Like
How to Amplify DevOps with DevSecOps
May 22, 2024Generative AI: Use Cases and Risks in 2024
May 29, 2024Smart Service Management
June 4, 2024