Pythia provides expert horse racing knowledge to prospective buyers, with accurate estimates on future performance of race horses

19 June 2024 – Pythia Sports is now using its substantial experience and resources in horse racing modeling to provide advisory services to interested parties looking to buy and sell bloodstock.

Pythia has so far advised on deals involving leading trainers such as Emmett Mullins, Hugo Palmer, John and Sean Quinn. All trainers and owners have trusted Pythia’s data-driven approach in their buying process.

Advertisement

Pythia Sports consumes data on races globally and builds predictive features that are then worked into sophisticated models, the output of which are sold to clients. It also provides pre-race and in-running odds, as well as risk management services to bookmakers and rights holders

The recent move to advise on sales has so far resulted in 14 horses being purchased using Pythia data. Of these, four have run, producing two winners.

Ran Amok won for Hugo Palmer at Chepstow on Monday 17 June after Royalty Bay won for John and Sean Quinn at Thirsk on 30 May. The horse also holds an entry for The Albany at Royal Ascot on Friday.

Pythia compiles ratings for clients via a number of different individual models, which then use machine learning to connect data from past sales with the future race performance of each individual horse. Pythia then joins the dots to identify and determine the most important indicators of horse quality.

The central model is supplemented by a series of specialised sub-models, which derive from information gathered at sale: biomechanics, time, and stride.

Advertisement

With the biomechanics model, the horse’s conformation and gait are evaluated by tracking specific points on the horse’s body. The time model combines multiple data sources to assess the speed, acceleration and style of the horse’s gallop.

Also, the group has an expert racing video analysis team that can process hundreds of horses on a stride-by-stride basis to a high level of precision, even in the limited time between breezing and auction.

On top of these three core ratings, the models also account for strength of breeders, consignors and pedigree based on their past actions and performance.

The central model incorporates the scores from these distinct sub-models and identifies in which instances each model excels, learning and improving as it develops.

As is the case with betting, Pythia can calculate an expected value based on the estimated probability of a horse’s performance and potential earnings.

Advertisement

Freddy Galliers, Director at Pythia Sports, said: “We are delighted to venture into the world of bloodstock, and we feel this will be a really exciting next stage of our journey as we continue to put our footprint on the horse racing industry.

“Our expertise lies in horse racing, making this a logical step for us. We don’t claim that our data is going to be the ‘silver bullet’ for bloodstock. However, we do believe that combining data-driven insights with traditional methods can provide a competitive edge. This approach can help differentiate us and our partners in the industry.

“The proof will be in the results, but we’ve already seen that horses we rated highly at the sales have gone on to win their maiden and novice races. We look forward to exploring further this section to see if there are new opportunities that can be delivered to the market.”