U of A Symposium: The Data is Out There. Now the Industry is Reaping its Benefits.

Technological changes on the immediate horizon for the racing industry won't be so much of the gee-whiz or sci-fi variety. Instead, they'll be more like much-needed ease-of-use and cumbersome-task-simplifying tools that are currently being built and driven by massive collections of data.

Tuesday's Global Symposium on Racing hosted by the University of Arizona Race Track Industry Program in Tucson kicked off with Lisa Lazarus, the chief executive officer of the Horseracing Integrity and Safety Authority (HISA), articulating a significant data point–that as of Dec. 10, 2024, the fatality rate of Thoroughbreds under the Authority's jurisdiction has dipped to .88 per 1,000 starts so far in 2024, a decrease from 1.23 fatalities per 1,000 starts one year ago on the same date.

Then, in a separate symposium discussion titled Artificial Intelligence (AI): Transformative Applications for Advancing Horse Racing, a group of experts from various fields underscored that while data is the building block, in order for it to be transformative, that information has to be put in a format where a computing system can sift and parse it properly. Then someone else (i.e. a software developer) must take that raw data output and present it to the end-user in a way that is useful.

That end-user could be a regulator like HISA, or other stakeholders, like owners, trainers, veterinarians, racing office workers, bettors, or even the buyers of young horses at auction.

Lazarus credited the Authority's growing database of 4 million veterinary records as being key to the decreased fatality rate.

“The first couple years of HISA were building the foundation. It was like triage,” Lazarus said during a solo presentation titled HISA in 2025: A Look Ahead.

“There was a lot to get done,” Lazarus continued. “We had to just get things up and running. But now we're at a point where we can take all of that work and everything that we've gotten from the industry, and give back to the industry in terms of information and products and really bringing the industry to the next level.”

In terms of HISA's rulemaking and operations, Lazarus said, “we can start to be a lot more intentional about matching the data to the regulations.”

That includes, Lazarus continued, “making sure we have regulations that are supported by the data and they make sense; that they are achieving the goal that we want, which is keeping horses and our riders safe.”

Mark Midland, the co-founder and chief executive officer of the online site Horse Racing Nation (HRN), concurred, then riffed more broadly on the topic.

“It really all starts with the data,” Midland said, explaining that once you have enough reliable data, you can build new tools to utilize it.

Midland noted that although HRN primarily functions as a Thoroughbred news site, about three years ago its leaders wanted to delve more into data and how to apply it handicapping–and beyond.

As examples of how HRN has branched out, Midland said his company has recently done data-driven studies on testing the accuracy of and automating the process of making morning-line odds (so that bettors can make better-informed choices), independently verifying veterinary studies of risk factors for injury (by inferring potential injuries based on non-veterinary pieces of data, like a horse's weight and speed figures), and by trying to predict the sales prices and eventual racing performance of young horses sold at auction.

“The more data the industry collects, the more it's going to be available for everyone for more learning going forward,” Midland said.

Kyle McDoniel, the president and chief operating officer of Equibase, moderated the AI panel. He gave a real-life example of the types of tools that industry participants want technology to deliver.

“When I think about the racing secretaries' meeting we had [Monday], one of the conversations that we had was around how veterinarians are getting frustrated because there's a lot of paperwork involved now,” McDoniel said. “There's a lot less time with hands on horses. So, to think about how AI [could] just get some of that administrative paperwork out of the way and [make] it more efficient [to] get more people time to do the jobs that they love and are really most important.”

Dr. Dionne Benson, the chief veterinary officer for The Stronach Group (TSG)'s 1/ST Racing, said that digitizing the daily minutiae of racing was the primary reason her employer developed the AI-driven product Racehorse 360.

The product, Benson said, was trialed for a year at the now-closed Golden Gate Fields before coming online at Santa Anita Park four months ago. Racehorse 360  has uses for veterinarians, racing officials and horsemen, and TSG plans to roll it out to other tracks in its corporate portfolio in the near future.

“Being the one who's out in the field, I just want my tablet. That's all I want to have to deal with,” Benson said, noting that the tool allows her easy access to looking up any of 50 separate injury-risk factors that might apply to any given horse in the database.

Benson said she can then do an exam and transmit the results to the proper racing officials or regulators, all while adding to the existing database of a horse's records, which can even include archived videos of the horse in motion for future health-check comparisons.

Lisa Lazarus | Nikki Sherman/Equi-Photo

Lazarus said HISA even monitors data on itself.

Lazarus said that it takes an average of 78 days from sample collection for HISA's equine drug-enforcement arm, the Horseracing Integrity and Welfare Unit (HIWU), to resolve a case.

“That's better than it used to be under most state systems. But we still want to work to improve that number,” Lazarus said.

In the near future, Lazarus said, “Instead of the labs reporting to HIWU and HIWU reporting to the veterinarians, there's going to be direct reporting from the labs to the regulatory veterinarians of results so that we can actually speed up this process. So that we improve, essentially, the time between sample collection and when the results come back, which is important.”

Michael Novak, a software developer with racing experience, wants to bring efficiency to owners and trainers via AI. In January 2025, he'll be launching a product called Backstretch that will be able to instantly scan through multiple condition books nationwide to find races that instantly match criteria selected by users.

“That's sort of the goal of a lot of software that I've been trying to build over the last 20 years, is really making software easier to use,” Novak said. “The experience about designing the interface and presenting exactly what the user needs when they need it is really powerful. And I think in terms of the software industry, you're always striving to make things easier. We don't always succeed at that. There's a lot of software that's very hard to use.”

Speaking relative to horseplayers, Midland said the idea of automating a morning line–like HRN has been doing for Colonial Downs–isn't to put morning-linemakers out of work. Rather, it's for a track to be able to put out the most accurate assessment of probable odds as possible to the betting public, which is a tool that smaller-scale horseplayers need in the current age of computer-assisted-wagering entities that dominate America's pari-mutuel pools.

“I think one of the problems with computer wagering today is the computer teams have all these tools, and the [smaller-scale] players have none,” Midland said, noting the lack of a “level playing field slanted against smaller bettors.

“Right now, I don't think it's a fair fight, and I think that's a big challenge to the game,” Midland said.

To try to gauge prices and future performance at horse auctions, Midland said the HRN data-crunchers recently fed 24 years of Keeneland sales information into a database, a process that took six weeks.

“We're not horse buyers [but] we're talking to people, trying to understand [how values get put on horses],” Midland said. “And it seems like the owners that I've talked to, when they're going to the Keeneland sale [where there are] 4,000 yearlings, it's hard to know where to focus and which ones to look at to short-list.”

Midland said, “We tried to undertake the question of performance. You know, 'Is this horse that's never raced going to be any good?' And there's a ton of data here.”

From a data set containing 7,500 yearlings sold at Keeneland in 2018-19 and following how those horses' racing careers turned out, Midland's team determined the highest-priced Thoroughbreds weren't always the best performers.

“We kind of see this as an inefficient market,” Midland said. “I guess that's probably not a total surprise to anyone. But it's a real opportunity, and we may come out with [yearling auction] 'draft picks.'”