Whether it’s quantitative data like ball trajectories, or qualitative data from scout observations, Major League Baseball (MLB) teams have plenty of valuable data sources. But how are teams like the Tampa Bay Rays coupling data with AI to enhance decision-making, optimize contract processes, and improve the odds of winning games?
This question was discussed in detail during the AI and machine learning (ML) “Panel at the Park” at Tropicana Field on Thursday, September 5, 2019. The event featured three key speakers: James Parks, AgileThought’s chief data scientist; Will Cousins, the director of baseball research and development for the Tampa Bay Rays; and John Higgins, the senior vice president of administration and general counsel for the Tampa Bay Rays.
Here’s a recap on some of the key takeaways:
Data is Imperative for the Tampa Bay Rays’ Success
Every piece of data—from box score statistics to merchandise sales—influences how the Tampa Bay Rays operate on and off the field. Off the field, data is crucial for budgeting purposes, Higgins said. By analyzing ticket sales and concession activities, the Tampa Bay Rays organization not only determines which vendors they should continue to work with, but also helps vendors identify top-selling items for games.
On the baseball side, data informs everything from recruitment to trades, Cousins said. “The job of my department here is to use all the information that we have available to us, to help support the decisions that the team has to make on the baseball side,” he said. “There are a lot of different [decisions]: What player do we draft? Do we make a particular trade, or do we not make a particular trade?”
There’s also an in-game strategy element in which data can be used to help implement strategies on the field and improve odds of winning. For example, the Tampa Bay Rays track every pitch that’s thrown in a major league game, so they can advise players on the pitchers’ release points and how to adapt as they change. In fact, almost all major and minor league parks now have radar systems that track the trajectory of the pitched ball and the exact trajectory of every hit, Cousins said. Every MLB game generates data, and the major league becomes a giant data-sharing network—generating tens of millions of data points every year.
So, what does all this data mean for the Tampa Bay Rays and their approach to machine learning? According to Cousins, it means that knowing how to operate machine learning tools is no longer a competitive advantage—it’s a necessity. The real competitive advantage is knowing how to integrate these tools with domain expertise—a practice that, Cousins said, has given the Tampa Bay Rays a chance to be more experimental and innovative with game strategies while helping players make better decisions on the field.
There’s Still a Huge Gap Between Research and AI & ML Applications
Despite the advancements on the baseball side, the administrative side is a little slower to catch up with machine learning, Higgins said. For example, all of the team’s contracts—whether it’s player contracts, employee non-disclosure agreements (NDAs) or vendor agreements—are reviewed and signed by Higgins and other key parties. It’s a manual process, he said, that can be time-consuming depending on how many individual contracts need to be signed. Although electronic signature tool DocuSign has helped expedite this process, Higgins said there’s a machine learning tool that has helped even more: IntelAgree, an application that helps enterprises automate and optimize the contract management process using machine learning. By interpreting contracts based on their key attributes and clauses, the application can reduce processes that take 4-5 hours down to 5-30 minutes—and Higgins said he’s hopeful that machine learning capabilities like this will help expand processes on the business side.
Parks shared similar thoughts on the AI and ML trends that he’s observing with AgileThought’s clients: “To me, there’s a big gap right now between research and application,” he said. “In my opinion, the application of AI and ML is pretty far behind the research—the things that are possible.” He added that there are plenty of tools, particularly with Microsoft, that can help companies go from zero to a working solution in as little as four weeks. However, many companies are still unsure about how to get started with the data they have available. “We see an opportunity where a lot of companies either have data and are unsure how to monetize it—how to wrangle that data and use it as an asset—and also companies that may not have data,” said Parks. It’s a market, he added, that AgileThought is appealing to with Predictive Analytics Discovery—a solution for analyzing data, statistically evaluating data quality, and delivering a proof-of-concept (PoC) in four weeks.
More Companies Are Recognizing the Need for Data Science
Although many companies are still struggling to identify their own machine learning use cases, most companies are recognizing the business value of data science. And the good news, Parks said, is that most organizations already have someone that’s ready to move into a data science role, provided that they have the right support. If an organization already has someone in a business intelligence (BI) or data analytics role, he explained, they just need to be given the right tools and knowledge necessary to help transition them into a data science role.
Furthermore, Parks said the attitude on using research to solve problems is changing. Historically, he said, companies didn’t use research to solve problems. But now, data scientists can tap into the data science community to extract research, analyze what worked and what didn’t, and apply those findings to current problems. It’s a practice, Parks said, that goes hand-in-hand with the iterative nature of agile development, which can help companies be more adaptive to business changes.
Want to learn more about machine learning and predictive analytics? Don’t forget to check out our blog for more industry insights, helpful tips and upcoming events.