AI Moneyball: One MLB team bets data will take it to the World Series

At Databricks’ annual summit in June, the data analytics and machine learning powerhouse touted its recent $1.3 billion acquisition of open-source AI startup MosaicML and showcased its industry impact with a parade of high-profile partners—from Microsoft to Condé Nast—who took the stage to sing Databricks’ praises.

In the lineup of corporate names, one stood out: the Texas Rangers, a (checks notes) professional baseball club from…that’s right, Texas.

We chatted with Alexander Booth, assistant director of research and development for the Rangers, at the summit, where he explained the club’s approach to data.

Baseball’s love affair with data was sparked by Moneyball (the book, authored by Michael Lewis, about Oakland Athletics GM Billy Beane, later dramatized in a movie starring Brad Pitt), Booth told Tech Brew. And AI is enabling a new generation of strategists and coaches to find the next Moneyball, he said.

The club uses AI for everything from game strategy and lineup decisions at the major league level to player development and pitch design in the minor leagues, as well as for drafting and trading players, he explained.

Traditional solutions, which include both data lakes (for unstructured data) and warehouses (for structured data), turned out to be expensive and inefficient for the quantity and type of data Booth and his team use.

“Our warehouse was fine when we were just looking at things like hits or discrete events,” he explained. “But when we start analyzing player positions, frame-by-frame data, biomechanics data, weather data, it’s really hard to do those transformations inside a warehouse.”

Databricks’ “lakehouse” model, which brings together both data lakes and warehouses, was the Rangers’ pick for “reformatting our modern data strategy” to provide “speed and agility in terms of data availability,” Booth told us.

Baseball has added a number of brand-new data sources to its toolkit in recent years, Booth explained. That includes ballpark weather data and biomechanical data, which he said spurred the Rangers to find a new cloud-based data processing solution.

The “technological explosion” in baseball isn’t restricted to MLB, he said.

High school games, for example, are also tracking pitches and player movements. “And that data then comes up to us at the major league level and we use that to analyze for the draft,” Booth explained. “What’s the likelihood that this amateur player will perform at the major league level in a few years?”

Booth pointed to the 2023 amateur draft this week, during which the Texas Rangers are relying on more data than they’ve ever had on high school and college players.

The supremacy of data has had an outsized (and some say detrimental) impact on the game of baseball, prompting the MLB to implement new rules to “mitigate some of the changes” fueled by analytics that made the game less engaging, Booth said.

As the industry juggles data-backed strategies and fan attention spans, Booth hopes data can take the Texas Rangers all the way.

“What I want to bring to the fans…is a World Series. And by using analytics, and machine learning predictions, and these innovations, and Databricks, we are trying to optimize the chances of winning the game as much as possible."