Mets' Robert Stock building AI-backed pitching analytics website

PORT ST. LUCIE, Fla. – The Mets are conducting a pitchers’ meeting and Nolan McLean has a question.

Pitchers want to limit “barrels," a stat that, at its core, indicates a well-struck ball that has a higher probability of falling for a hit. Pitchers also want to limit walks. But McLean wonders: Wouldn’t pitchers who limit walks by pounding the strike zone also allow more barrels?

A voice chimes in, and it’s not pitching coach Justin Willard, or the other experts the Mets have on site, tracking pitches, overlooking bullpens, and compiling and analyzing data, their various tripods and cameras and video screens in tow.

It’s veteran reliever Robert Stock.

McLean, 24, is primed to be the next big thing. By his own admission, Stock, 36, “has had extremely limited MLB success.” After the Mets signed him in November, he tweeted: "And yet MLB teams continue to give me chances. Why? Because I continually find new ways to improve.”

On this day, Stock pulls up the database on his phone. No, the two stats aren’t correlated, he tells McLean. And with further thought, it makes sense: pitchers with command issues may be working their way into hitter’s counts, and that can lead to damage as much as it can lead to walks.

Stock simply imported the question into his pitching analytics platform, “Stockyard Baseball Co.,” a database he built with the help of artificial intelligence.

“Year after year, I watched people on Twitter or wherever put out very interesting analytics and data and I always wished I had a database where I could do that,” he told Newsday last week at Clover Park. “For instance, teams can tell you what your grades are when you're inside an organization, but it's harder to find out: What if I changed my release point by an inch higher or lower, what would my grade be?... You can ask the pitch model, ‘Does this sound like a good idea?’ ”

A pitch grade is a method of assessing the quality and potential of, say, a pitcher's slider.

Stock’s website is live but still in its testing phase, and it works by marrying MLB’s pitching data (over 8.9 million pitches) with “machine learning” – a subset of AI that uses algorithms and available information to get smarter, identify patterns and draw conclusions. Because the data sets are public on sites like Baseball Savant, it’s easier to compile that if you were starting from scratch.

It’s also part of a growing trend. With so much information at their disposal, some players are taking a hands-on approach to figuring out what, exactly, makes them tick.

Retired Mets reliever Trevor May, who spent nine seasons in the big leagues, is similarly building a data-driven baseball analytics platform.

“I think as a player, you have to be curious about the game,” said Mets starter Clay Holmes, whose own penchant for analytics-based tweaking reinvigorated his career a few years ago. “You have to be open to things. There are principles of the game that's always going to be there, but there’s always new ways to go about them …You see guys like Stock, it's cool just to connect and hear about it. And I think that love and joy that it brings is infectious and it pushes other people too.”

For guys like Stock and May, it’s not about replacing the carefully curated [and sometimes mysterious] pitching models, but getting information into the hands of the people who need it the most, doing so transparently, and answering questions that maybe others haven’t thought to ask.

Stock, who threw a no-hitter for the Long Island Ducks in 2023, is fascinated by what he calls “stickiness” - stats that follow a pitcher year to year, and are thus more “skill-based and predictable” as opposed to context driven. May, meanwhile, has developed a metric called “misfire,” which seeks to measure intention when it comes to command, such as pitchers who miss on purpose versus those who throw pitches that begin out of the zone and break toward the plate.

Outside of the “truly elite guys” baseball is populated with “the guys like me where you’re kind of fringe, or you’re a reliever and you need something unlocked,” said May in a phone interview. Besides the top-tier players “95% [of us] are like, ‘I don't want to go back to the minors ever. How do I do that?’ That is what I've always been concerned with. And I think if you take ownership of your own relationship with analytics and … if you’re just a little bit curious about the context [of why things happen] you can find little, tiny things that you can slightly adjust that will just completely fix a big problem.”

There are ethical and practical issues when it comes to how and why AI is used, but it’s clear Stock views it as a tool rather than a catch-all. Both he and May are also aware of the pitfalls as it pertains to accuracy. Stock’s 14 years of professional pitching helps him see what conclusions pass the sniff test, but he’s solely relying on that to filter out “AI slop." He also crowdsources feedback to hone the process. May, who retired in 2023, uses personal experience and outside guidance to test for accuracy.

“That’s the scientific method,” May said. “You’re trying to prove yourself wrong…

Sometimes, “your red flags go off,” he added. “So you just run tests. [You give it to] 20 different people and see them do it individually, and see if it comes out correctly. If you get 20 out of 20 right, you're pretty confident the rest was done, right, too. That’s the stuff you have to make sure you're paying attention to, because it could very easily be giving you something that, not for any fault of the process or anything [isn’t accurate]. It just didn't fully understand what you wanted, and you’ve got to make sure that it explains it in a way that makes sense to you.”

May is exploring questions that nagged at him during his career and noted how, in 2019, then Twins pitching coach Wes Johnson used Hawk-Eye 3D, a system that provides 3-D pitch-tracking and biometric data, to help him identify a small fix that upped his velocity.

“That was the inspiration,” May said. “It changed my career.”