Data Scientists on the Sidelines

The influence of data science in the sports industry could be a game-changer for franchises willing to think outside of the box.

No matter how you look at it, the sports industry is a numbers game--wins and loses, strikes and fouls, and ultimately player and team performance statistics. So, in a world that’s powered by data, where are all of the data geeks on the sidelines?

Though it may be awhile before data scientists are sought after as draft picks, franchises around the globe are at the intersection of exploring how data can fuel performance on and off the field of play. From partnerships with startups, to front office positions, data and AI experts are quickly making their way into the 73.5 billion dollar sports industry.

Stats, a Chicago-based data and technology company is among the pioneers of sports data and analytics. Since 1981, the company has been gathering information from every corner of the sports industry to improve the accuracy and reporting of data and statistics, even placing employees in the stands to code games in real time. The company, which is already using advanced technologies, is now growing its artificial intelligence team to provide a more sophisticated approach to data delivery.

“Artificial intelligence can gather and process even broader data,” said Richard Henderson, Chief Revenue Officer for Stats. “It can provide predictive analytics to coaches, showing them, for instance, a play in which their players failed to score and recommend a different run that could result in a goal.”

While Stats, along with other teams of AI experts have clear visions for the use of machine learning and computer vision in the sports industry, the most common setbacks may not be related to technology at all. Instead, variables in human performance and behavior could spark algorithm deficiencies that are too complex for even the most innovative AI solutions.

Dr. Joel Sokol, Associate Professor and Director of the Master of Science in Analytics at Georgia Institute of Technology, explored potential challenges when measuring and predicting data that relies on aspects of human performance. “The difficulty isn't necessarily in the AI or robotics, but in our knowledge of how to best define the problem,” said Sokol. “For example, AI has gotten really good at visual things (such as facial recognition) because we're good at defining the issue -- specifically, for each pixel, what's its color, brightness, etc.”

The difference, as Sokol explained, is deep within the data. “Once we give the AI the right data, it is very good at determining the right answers. But in sports, important issues like day-to-day variability aren't yet well-defined on the input side,” he added.

In addition to exploring the use of AI’s machine learning and predictability capabilities, Sokol, along with others, have suggested that sports franchises take an interdisciplinary approach when attempting to decode uncertainties and variables between data and human science.

When Dr. Lorena Martin joined the Seattle Mariners in 2017, she was recruited to do just that. As the team’s first-ever Director of of High Performance, her role was created with data science in mind and she was tasked with integrating her expertise in physiology, psychology, sports science, and statistical analysis to pursue maximum player and team performance. In an interview published by the Mariner’s, Martin shared her data-centric approach to understanding when and how athletes are at peak performance levels.

Each player has their own formula for peak performance, Martin explained, and each equation needs to be individualized in order for them to perform at their best. “It’s crucial at this level to customize [the data] for each player because they are completely different, not only in style, but in player position, and body composition,” she said.

Although Martin’s holistic approach begins to address the influence of human behavior on data-driven solutions, there are still many uncertainties regarding the marriage of science and sports. Over the years, data and analytics have provided insight on many aspects of the game, but it’s probably safe to say that coaches and players aren’t ready to take direction from robots anytime soon.

Dr. Konstantinos Pelechrinis, Associate Professor at the School of Computing and Information at the University of Pittsburgh, has always had an interest in merging sports and data science, but he is realistic about the challenges that both data scientists and sports franchises will face as the two worlds collide.

“I don’t think there are many differences in principle between sports data and other types of data, but some data challenges can be more pronounced in sports data,” said Pelechrinis. “One example would be collinearities when it comes to analyzing lineups for dividing credit to players in the lineup. While this is a problem that can be present in many types of data, it is very pronounced in sports data.”

From his point of view, the trend of incorporating data science into front office positions will continue to grow, but with the anticipation that conflict may arise when it comes to implementing changes based on data.

“The use of data and analytics can provide different angles to decision making and the challenge is having the ‘traditionalists’ and the ‘analysts’ co-exist and understand what each other is bringing on the table,” said Pelechrinis. “One thing we have to remember is that statistics and data have always been a part of sports. For example, NFL teams identify the ‘tendencies’ of their opponents and data scientists do the same, it’s just that we call them ‘patterns’.”

Perhaps, as franchises continue to partner with data scientists--whether through a front office position or otherwise--experts of varied backgrounds will find new ways to improve access to data and AI technologies in order to truly make data-driven changes.

“Everyone will need to keep an open mind,” added Pelechrinis.“The power of inertia can be much stronger in the sports industry as compared to other industries, but it’s important for everyone to understand that the data is not only valuable for improving player and team performance, but also in understanding the fans, ticket prices, and team promotion.”

About the author

Jennifer Palma Sanchez is a communications expert with over a decade of experience in marketing, public relations, and corporate communications. Follow @palma_lita to keep up with her perpetual desire to learn something new.