• Hi Guest Just in case you were not aware I wanted to highlight that you can now get a free 7 day trial of Horseracebase here.
    We have a lot of members who are existing users of Horseracebase so help is always available if needed, as well as dedicated section of the fourm here.
    Best Wishes
    AR

Can AI Find Winners in UK Flat Racing?

cpw

cpw

Yearling
With the rise of AI, we’re all curious—can it really help us pick the winning horses? 🤖🐎

AI has transformed many industries, but can it crack the code of UK Flat Racing? Analysing data like race history, track conditions, and horse form allows it to offer insights traditional methods might miss.

Do you think AI could give us an edge I have been exploring for a while? Let’s discuss! Share your thoughts 👇
 
Regretfully i cannot contribute Re AI because i do not use it but i see we have members who do so hopefully your thread will attract some interest. Reading your own thoughts on the topic should increase the chances. :)
 
I have been exploring and have been surprised how easily AI pulls all of the information together. You can simply cut and paste the information and then ask away.
  • Data Integration: It can pull in and process vast amounts of historical data on horses, jockeys, trainers, and races. This includes horse performance in different conditions (e.g., track surface, distance), jockey win rates, trainer success at specific courses, and more.
  • Pattern Recognition: By analysing previous race results, ChatGPT identifies patterns—such as how certain horses perform under specific track conditions, or how jockey changes impact results. These patterns, which may be too subtle for humans to easily spot, help the AI make more informed predictions.
  • Contextual Understanding: ChatGPT, Claude, Gemini can dynamically assess real-time data such as last-minute changes in weather or odds movements, adjusting its predictions accordingly. For example, it might identify that a certain horse performs well in wet conditions and adjust its prediction if rain is forecasted on race day.
  • Probability Estimation: After processing all relevant variables, the AI assigns probabilities to each horse’s chances of winning based on historical data and current race factors. These probability estimates give users a clearer idea of which horses have the best potential for success.
 
What you will soon discover is LLMs are hopeless at Maths. Which is to be expected I guess since it's a LARGE LANGUAGE MODEL :) The thing most people don't realise is you can use Google's BIG QUERY and combine with AI. Also with the advent of plugins you can basically connect APIs right into your AI client.
 
It's early days but guys it will save us so much time. I am combining my old methods and then feeding it a final report which I can just query. So have any of you explored?
 
cpw cpw your above looks interesting but i suspect that Bookmakers and some high stakes punters have already been using AI and if simular questions and data are involved is or will this affect the odds. ? For myself making a profit from this game is all about obtaining value prices against an expected winner and i have found that to achieve this some contrarian or minority thinking is a need.
 
Here is an example. I have weighted my factors using a Google's BIG QUERY's data platform then used it's neurel net to provide me with weights for each winning horse in the past from. Then get the AI to apply each factor to a race running today. The AI produces output like this

1727084225843.png
 
mick said:
cpw cpw your above looks interesting but i suspect that Bookmakers and some high stakes punters have already been using AI and if simular questions and data are involved is or will this affect the odds. ? For myself making a profit from this game is all about obtaining value prices against an expected winner and i have found that to achieve this some contrarian or minority thinking is a need.
Click to expand...
I suspect not many have the time or skill to do what is necessary. What you have is knowledge that many don't have. Although many have access to AI they don't know how to apply it. Think of it like having Excel at your disposal but much more. You can ask questions and get answers back that make sense.
 
cpw cpw the 5.15 Hamilton you mention is my race of interest today and i have played at a rare for myself short price. Best of for Angel Amadea if you are backing her.

Mon 23rd Ham 5.15 : Gentle Ellen @ 10/3

A 6 fur 0-60 with 7 run. I suspect some gaming involved Re her three most recent runs with 3lto a 0-78, 2lto fast ground and lto a trip too far and these have bought a begrudging 5lb OR reduction from the BHA which creates the chance here. She has won over this CD has proved her weight carrying ability and likes some give in the ground.

Her all in 15% 3-20 S/r improves to 1-1 in this 0-60 grade and she looks placed to win dare i say a little plot. The stable appear in good nick this month and i have no problems Re the rider, and both produce above there avg S/r at Hamilton.

B365 and others are 10/3 Bog and this just about enables becoming involved.
 
mick said:
cpw cpw the 5.15 Hamilton you mention is my race of interest today and i have played at a rare for myself short price. Best of for Angel Amadea if you are backing her.

Mon 23rd Ham 5.15 : Gentle Ellen @ 10/3

A 6 fur 0-60 with 7 run. I suspect some gaming involved Re her three most recent runs with 3lto a 0-78, 2lto fast ground and lto a trip too far and these have bought a begrudging 5lb OR reduction from the BHA which creates the chance here. She has won over this CD has proved her weight carrying ability and likes some give in the ground.

Her all in 15% 3-20 S/r improves to 1-1 in this 0-60 grade and she looks placed to win dare i say a little plot. The stable appear in good nick this month and i have no problems Re the rider, and both produce above there avg S/r at Hamilton.

B365 and others are 10/3 Bog and this just about enables becoming involved.
Click to expand...
My AI has this. 1727085099600.png
 
Would be interesting putting gentle ellen and havana pursuit as i think this one could be big danger only not my tupe of race as nothing sticks out the worries with havana are the draw the no track exsperience and i have new way of applying horses and dosnt fit it this race.
 
The one that does stick out to me is strong johnston in the 4 08 i see some money came for it and i think will have big chance today.
Knocking on the door of this mark and i just wish was drawn in middle but will go close i feel.
 
Hi cpw cpw

Mark Littlewood has been using AI for some years now and you might find like minded people to interact with

Improve Your Betting With SmarterSig

It is ok using AI to predict winners but if backing the predicted winner blindly it will not necessarily result in profit if you are not backing those whose odds are better than your AI predicts their probability

mick mick has been appraising the odds for decades and knows When the Bookies have made a mistake with their odds. He knows from experience that there will be losing runs but because he is Getting value odds he will be in the Black by the end of his financial year.

Bill Benter used machine learning to make a small fortune, but he only managed that because he was based in Hong Kong and had the Tote odds. Being fed into his computer. Until he realised that the crowd were actually good at predicting winners and used that information, his programme was losing money

see

Bloomberg - Are you a robot?
 
Last edited:
Chesham said:
Hi cpw cpw

Mark Littlewood has been using AI for some years now and you might find like minded people to interact with

Improve Your Betting With SmarterSig

It is ok using AI to predict winners but if backing the predicted winner blindly it will not necessarily result in profit if you are not backing those whose odds are better than your AI predicts their probability

mick mick has been appraising the odds for decades and knows When the Bookies have made a mistake with their odds. He knows from experience that there will be losing runs but because he is Getting value odds he will be in the Black by the end of his financial year.

Bill Benter used machine learning to make a small fortune, but he only managed that because he was based in Hong Kong and had the Tote odds. Being fed into his computer. Until he realised that the crowd were actually good at predicting winners and used that information, his programme was losing money

see

Bloomberg - Are you a robot?
Click to expand...
Hi Chesham Chesham

Thanks for the feedback. Yes, I know Mark very well from that group. The work that group did was amazing. I miss the original printed format. I was part of the original SmartSig group many years ago before SmarterSig. The AI we used back then was NN and I still use it as a neural net. What we have now is a completely different animal. Yes, very familiar with Bill's work I didn't think he was using AI. There is confusion between machine learning and large language models. MySportsAI would lose most people.
 
Many AI systems can generate code for you, but you'll typically need to download and run it (e.g., with Python, ensuring you have the latest version). However, not all AI-generated code is error-free. In my opinion, AI is evolving rapidly, and finding the best one for developing coding solutions, especially for horseracing selection systems, can be challenging. One of the most time-consuming tasks for me is sifting through data, and AI should soon be able to significantly streamline this process.
 
Paul Brown said:
Many AI systems can generate code for you, but you'll typically need to download and run it (e.g., with Python, ensuring you have the latest version). However, not all AI-generated code is error-free. In my opinion, AI is evolving rapidly, and finding the best one for developing coding solutions, especially for horseracing selection systems, can be challenging. One of the most time-consuming tasks for me is sifting through data, and AI should soon be able to significantly streamline this process.
Click to expand...
That's my point Paul. We are thinking old school. We don't need to download any code and run. Just provide context and data and it does the rest. You simply have to ask the questions. You have an SME and Data Analyst all built in. You just ask it about the data it has in plain English :dreads:
 
You must log in or register to reply here.
Back
Top