Whan analysing horses, Pythia Bloodstock measures three categories for its model: time, stride and biomechanics. Getty Images
Whan analysing horses, Pythia Bloodstock measures three categories for its model: time, stride and biomechanics. Getty Images
Whan analysing horses, Pythia Bloodstock measures three categories for its model: time, stride and biomechanics. Getty Images
Whan analysing horses, Pythia Bloodstock measures three categories for its model: time, stride and biomechanics. Getty Images

From gut instinct to algorithms: how AI is changing horse racing


Add as a preferred source on Google
  • Play/Pause English
  • Play/Pause Arabic
Bookmark

It is a bright and breezy opening day to the Flat racing season at Newmarket and Michael Owen is casting his eye over his runners near the pre-parade ring and liaising with his trainer Hugo Palmer.

Owen, the former Liverpool and England striker turned racehorse owner with his own high-end stables in north west England, has hotfooted it to the course from the famous Tattersalls sales ring where, every year, hundreds of millions of pounds are spent chasing the dream of buying a potential superstar.

There, he had been running the rule over potential purchases for himself aided and abetted by relatively new AI-rooted technology and algorithms which provide invaluable data and insights to assist the decision-making process.

“It's enhanced the way we do things and we embrace it,” Owen, 46, and who won 89 caps fir his country, tells The National. “It's great to cross-reference horses that may arrive on our wish list.

“Once we look at the data behind it, it can either make us fall in love with a horse, or even more it can warn us off, which is in many ways equally as important as trying to find a good horse.

“We don't want to be wasting our bullets. Every bullet that we fire, we want it to have dotted all the I’s and crossed all the T’s. So, if we can fine-tune that process, then all the better. “

The technology has been developed by the London-based company Pythia Bloodstock. It films horses at Breeze Up sales, which is where unraced two-year-olds gallop for two furlongs in front of prospective buyers.

The model

It measures three categories for its model: time, stride and biomechanics. These are scored individually against a baseline of a mass of data to produce a score for each. These are then combined to offer a percentage likelihood of a horse’s prospective capabilities and whether it might be rated at more than 90, the official rating, which denotes a horse of considerable ability.

Steve Davison is Pythia’s is head of commercial operations. “AI is obviously a big buzzword, and it gets a lot of pick up, but we refer it as machine learning models. Each year we've got a brand new bit of data and our models are learning what we got right or wrong.

“As an organisation we go all the back to the 1980s. That puts us ahead of the game. We are used to gleaning data, sifting out the noise in it. AI is not a silver bullet for racing. You can't just drop all the time ratings and then say. But our technology can and does work. More than 20 horses have been bought using our models and we have generated interest from many jurisdictions, including Australia.”

All those involved concur that it is not designed to replace the human eye or the expertise of the horseman, nor can it offer the insights that an equine veterinary expert can provide. Neither can it "get between a horse’s ears" says Billy Jackson Stops, the leading bloodstock agent, recognised officially as the world’s best last year.

Jackson Stops has sourced horses for Owen as well as many other clients, and is also the racing advisor to Sheikh Nasser bin Hamad Al Khalifa of Bahrain, a country fast playing catch-up on the international racing stage.

“No AI in the world is going tell you what goes on between a horse’s ears. The overall and resounding feature of any decent horse it that it has got a good temperament, good brain, and a will to win.”

Game changer

That said, he is a keen advocate of Pythia’s technology believing in time it could become a real game changer. “You still must make your own personal judgment, but the data is a very, very good aid to that. By and large it's going to improve your chances of buying a better horse. I think it could be massive for racing. It is going to get bigger and better every year as they refine it.”

He says, above all, it is a huge asset to finding horses that potentially could be undervalued.

“If you're coming into a sale with a kind of middle order with regard to what you can spend, this is where you're find value. And you're always looking for the gem that costs a few quid as opposed to seven figures.”

Last year was a good one for Jackson Stops who had four Group 1 winners across the world from the horses he had bought.

One of those hidden gems was Gewan, a horse he bought for only £80,000 and who proved to be so good as a two-year-old that he was one of the favourites for this season’s 2,000 Guineas. Tragically he died earlier this month after an accident in a racecourse gallop.

“I am data-driven and when I buy it's all down to numbers. I'm a big believer in stride length for example. That demonstrates how efficient a horse is. It uses everything better. It uses its oxygen better; it breathes in better. It can then go longer. It's obviously more aerodynamic and uses itself better.“

Former England footballer Michael Owen says horse racing lags behind other sports when it comes to using data. PA
Former England footballer Michael Owen says horse racing lags behind other sports when it comes to using data. PA

Owen knew that his life after football would involve horses. He bought Manor House Stables when he was 23 and started the process of establishing a leading training centre for which he hired Palmer to be his trainer four years ago.

When he started out, he was staggered by the paucity of data available in racing, which like football, he said, lagged far behind other sports.

“Racing has been pretty slow to adapt. So too was football. Both are catching up but are still behind athletes like rowers, runners, cyclists and Olympians who have been at the cutting edge for probably the last 20 or 30 years.

“I have been in the racing game pretty much all my life and I used to be astounded how much money spent on very little data, on a whim almost.

“When you turn up at a sale and you look at a thousand horses and you get 15 seconds with them, they walk up and down and then you decide whether you're going to spend £100,000 or £200,000 on a horse.

“That to me just feels a little bit outdated. Pythia do a pretty unique and clever thing with their data. I feel as if it gives us a lot more comfort and a lot more back up. If we like what we see and the data supports this, you get two thumbs up. That then gives me a lot more confidence to get my wallet out of my pocket!

“I do think we're ahead of the curve trying to buy horses. I think our results on the racetrack would bear that out. And listen, it's always about trying to stay one step ahead of your competition.

“You've always got to keep pushing the boundary. If you sit still in this game or in any game or in any industry, you're going to be behind the times. So, it's important to have an open mind. And that's certainly what we've got.”

Owen’s day at Newmarket ended on a high. Not only did he have a winner, but he and Palmer later landed a filly at the sales which they had long admired. The unnamed daughter of St Mark’s Basilica cost almost £200,000.

“We’re thrilled to buy her,” says Palmer. “We thought she was a cracking individual. She performed really well on all the Pythia numbers, which is what we've been working on for a number of Breeze Up sales now. She was one of the ones we really, really wanted.

“She is the most expensive filly we've bought. We've had lots of 90-plus horses, but let's hope that she can be a real star.”

Updated: April 27, 2026, 4:44 AM