OddsOracle - AI-Powered Horse Racing Analysis
Advanced horse racing analysis powered by AI. Get detailed insights and professional analysis for today's races.
Top Tipsters ?
- Thread starter tacker
- Start date
trecelyn
Filly
I see a plethora of AI generated form analysis options out there at the moment. My understanding is that the response you get is down to the questions you ask and the data you ask it to get its selections from. All the models I see being pushed by platforms claim to be following the form. If every AI tool out there is looking at the form then they are getting their information from the same pool and come up with very similar answers.
My reckoning is the old pen and paper users who possibly think outside the box a little may suddenly find themselves with a bit of an edge by not following the AI revolution however attractive these ads make it sound
My reckoning is the old pen and paper users who possibly think outside the box a little may suddenly find themselves with a bit of an edge by not following the AI revolution however attractive these ads make it sound
AJ the Hobbyist
Filly
Race 2 – 2.05 Pontefract
Pre-race V15:
Win Pick: Instant Bond
Forecast Combo: Instant Bond → Melissa Honey / Invincible Ruby
Official result:
1st Instant Bond
2nd Melissa Honey
3rd Invincible Ruby
V15 Win Pick: WON
Exacta: LANDED
TOTE Exacta: £70.40 (P/L: +£68.40)
Boxed Trifecta: LANDED
TOTE Trifecta: £264.10 (P/L: +£258.10)
Instant Bond was an unlikely winner who, unusually, took an early lead, controlled the pace, and then outstayed the other runners. Did I back him? NO.
AJ - Good follow on for win picks, but I missed INSTANT BOND 20's on BFEX. WGAT
V15 'obby - That was the card’s sharpest lesson.
INSTANT BOND was the pure AU nerve-hold: weak market, cold jockey, confidence reduced — but still the uploaded points leader and still won at a proper price.
You were right to keep the wider discipline clean, though. Missing the BFEX 20s hurts, but chasing every exchange spike is how the structure gets bent.
Carry-forward is simple:
BFEX weakness is a caution for staking, not automatic rejection.
If AU is clearly dominant and the price is drifting, the horse stays structurally live — but exposure stays controlled.
WGAT.
Last edited:
trecelyn
Filly
I think you are missing the point I am trying to make, I am not saying AI does not work as I think you are trying to show with your example above that it does. There seems to a mad rush by people to use AI to automate all parts of their lives and if thats what they want absolutely fine. However there are shall we say charlatans and rogues out there who have seen an opportunity to train a model to identify the winners of certain races and with no knowledge or very little at the best they are using the most common winning factors to get their selections. Dare I say a lot of them will have visited forums like this one and harvested the best ideas to use with their software.
Once they have their model they peddle their wares to the masses through social media in some cases actually pointing out if you dont use AI you cant win. They have their initial offers, follow it up with the hook and then the big sales pitch. This is the part I was trying to draw attention to, the more of these "experts" there are, the shorter the odds at the front of the market where nearly all AI systems operate and possibly, just possibly the backers who use their own methods and maybe use information in a different way will find the odds more in their favour.
I may be totally wide of the mark but one thing I have learned if you follow the crowd in this game, the bookies will alter the odds to reflect this and you normally lose in the long run. My post was to encourage a bit of debate not have an AI breakdown of the race but therein may lay my answer, people are heading towards using AI for everything and are losing the ability to debate and offer opinion.
trecelyn
Filly
In the horse betting world I have seen some wonderful output by some of the members on here. AI is in its infancy and I think it will be a tremendous asset in some spheres going forward, but unfortunately it is currently being marketed as the answer to everything and is open to errors because some people use it to make their lives easier and do not check the output they get, then this gets compounded the more they use it as AI now thinks it is correct and deeper flaws start to materialise.I wondertrecelyn if you have seen anything AI that has impressed you?
Not me, yet.
I do a lot of writing, not pushing anything unless you want to read about mental wellbeing and the role therapeutic horticulture can play, but everything I write, I check, edit, reread, edit again and only when I am finally happy do I publish. With AI, and I can show you examples of it gone wrong, more often than not it goes published unverified.
I use AI for image generation so I am not totally against itbut to answer the original question, I am still looking but believe in the future I will find something that impresses
JennyK
Gelding
T
tacker. You obviously have some experience of AI and others may well have even more. There is a question re data that I can't answer through my database and Excel coding, and I am wondering if AI could, because if so it would be very useful and I'd be happy spending time with an AI application to get it to do it for me.
I use a very simple ability rating - total win Flat prize money divided by number of Flat races won. For horses who have only won in the UK, my Excel coding does it almost instantly for a field. But if any runner has won abroad, that performance isn't in my base and therefore necessarily my coding can't deliver the proper rating.
My question is, is it possible to instruct/programme (not sure of the right word) an AI application to consult the Post's database for each runner and make the proper calculation, ie include any overseas wins, and if so once one has input the race venue and time, how quickly would the list of ratings be available? I specify the Post because I know of no other free database which includes (as far as I can tell) all overseas wins for horses currently running in the UK.
I use a very simple ability rating - total win Flat prize money divided by number of Flat races won. For horses who have only won in the UK, my Excel coding does it almost instantly for a field. But if any runner has won abroad, that performance isn't in my base and therefore necessarily my coding can't deliver the proper rating.
My question is, is it possible to instruct/programme (not sure of the right word) an AI application to consult the Post's database for each runner and make the proper calculation, ie include any overseas wins, and if so once one has input the race venue and time, how quickly would the list of ratings be available? I specify the Post because I know of no other free database which includes (as far as I can tell) all overseas wins for horses currently running in the UK.
JennyK My understanding is you need something like a SUPER PROMPT but suspect you'll need to spoon feed AI anyway.
You should ask AI what the best way to ask it the question you want an answer to, make sure to tell it not to hallucinate or make things up.
I speak to people who run a solicitors firm and they show me some remarkable stuff that effectively will cost jobs within their company, on saturday i ask what's to stop AI from undermining the need for other solicitor firms deciding they perhaps won't need to use their company in the future ?
Not sure they answered.
They are forever seeking advice from CLAUDE ( whatever the latest version is) what to back in a particular race and there's the worry, a daft question but AI still gives them an answer and they are happy to swallow it.
Finally ask AI if bookmakers use AI and from memory the answer i got was yes but on a different level.
Leodis
Gelding
AI now is very much like the Betfair model was in its infancy. The few who got on board early, made a killing against the novices until the day came when the novices got equally good at it. Now it is almost impossible to find an edge with Betfair.
My advice, fill your boots with AI while you can.
My advice, fill your boots with AI while you can.
Leodis
Gelding
A lot of companies claim to use AI only to attract punters. Most of it is BS!
Leodis
Gelding
You could ask AI, but it's not guaranteed to come back with the correct info,
JennyK . That is if you are expecting it to behave like a web crawler or such.Try a simple prompt to return the overseas earnings/runs of a sample horse (you can check against) and see what it comes up with. Add some helper prompts along the way if it ain't quite getting it right.
JennyK
Gelding
T
tacker,
Leodis
Thanks. It sounds to me as if I'm better off doing what I do now - manually checking the Post site for the ability rating in the small number of situations where I could go wrong if the rating is much under-valued by missing a big win in (usually) Ireland, Bahrain or Meydan.
Fortunately there aren't many where it matters and if one specialises in a particular subset (sprint handicaps in my case) one mostly learns the ones who have had big overseas wins.
The great thing about Excel is that, with proper programming, the data extracted is always flawless; the only problem being, as with that needed to compute some ability ratings, not having the data in the base in the first place. I suppose Excel was a Microsoft team creation, but if it was one person's brainchild and I knew his or her name, they would be up there with VDW as pillars of my race analysis activity. I cannot imagine how long it would take to analyse a large-field handicap like the Wokingham in the way folk had to pre the home pc.
LeodisThanks. It sounds to me as if I'm better off doing what I do now - manually checking the Post site for the ability rating in the small number of situations where I could go wrong if the rating is much under-valued by missing a big win in (usually) Ireland, Bahrain or Meydan.
Fortunately there aren't many where it matters and if one specialises in a particular subset (sprint handicaps in my case) one mostly learns the ones who have had big overseas wins.
The great thing about Excel is that, with proper programming, the data extracted is always flawless; the only problem being, as with that needed to compute some ability ratings, not having the data in the base in the first place. I suppose Excel was a Microsoft team creation, but if it was one person's brainchild and I knew his or her name, they would be up there with VDW as pillars of my race analysis activity. I cannot imagine how long it would take to analyse a large-field handicap like the Wokingham in the way folk had to pre the home pc.
T tacker,
Thanks. It sounds to me as if I'm better off doing what I do now - manually checking the Post site for the ability rating in the small number of situations where I could go wrong if the rating is much under-valued by missing a big win in (usually) Ireland, Bahrain or Meydan.
Fortunately there aren't many where it matters and if one specialises in a particular subset (sprint handicaps in my case) one mostly learns the ones who have had big overseas wins.
The great thing about Excel is that, with proper programming, the data extracted is always flawless; the only problem being, as with that needed to compute some ability ratings, not having the data in the base in the first place. I suppose Excel was a Microsoft team creation, but if it was one person's brainchild and I knew his or her name, they would be up there with VDW as pillars of my race analysis activity. I cannot imagine how long it would take to analyse a large-field handicap like the Wokingham in the way folk had to pre the home pc.
JennyK I don't want to give the impression that AI lacks value, it certainly has plenty and keeps improving, it's just that i prefer to use it in a more educational and analytical way rather than a shortcut to picking winners.
Asked it a question about ascot and the draw bias.
Here’s the distance‑by‑distance breakdown of Ascot track bias you asked for, David — the version punters actually use, not the watered‑down “official” line. Ascot is one of the most distance‑sensitive tracks in Britain when it comes to draw and pace bias, especially on the straight course.
5 furlongs (straight) — The most draw‑sensitive trip at Ascot
Takeaway:
High draws (stands‑side) have historically held a strong edge on fast ground, but the bias can flip on soft.
Why:
- The stands‑side rail often rides quicker in dry spells.
- Horses drawn high can grab the rail and get a “free” running line.
- On soft ground, the far side can drain better, giving low draws the edge.
Practical betting angle:
- Fast ground → favour high draws
- Soft ground → check early races; bias often flips low
- Pace clusters matter massively — if all the speed is low, the bias becomes low.
6 furlongs (straight) — Similar to 5f but slightly less extreme
Takeaway:
Bias still exists, but pace distribution becomes more important than pure draw.
Why:
- The extra furlong gives horses more time to cross over.
- Big-field handicaps (Wokingham, Buckingham Palace) exaggerate whichever side is quicker.
Practical angle:
- High draws still preferred on good/firm, but not as dominant as 5f.
- If the field splits into two groups, the side with the stronger pace wins regardless of draw.
7 furlongs (straight) — The most tactically tricky trip at Ascot
Takeaway:
This is the least predictable of the straight-course distances.
Bias varies wildly by:
- ground
- pace
- field size
- rail placement
Why:
- 7f is long enough for horses to switch sides.
- The field often splits into 2–3 groups, creating “false” biases.
- The middle can be a graveyard if both sides have pace.
Practical angle:
- Don’t over-commit to draw bias at 7f.
- Focus on pace maps — the side with the most early speed usually dominates.
1 mile (straight) — The Royal Hunt Cup trip
Takeaway:
The straight mile is hugely draw‑dependent in big fields, but the bias changes year to year.
Typical patterns:
- Fast ground → high draws often dominate
- Soft ground → low draws or middle can be favoured
- Horses drawn in the middle can get trapped with no group to follow.
Why:
- The straight mile is long enough for ground differences to matter more than pace.
- The stands‑side rail can be a “golden highway” on firm ground.
- On soft, the far side often drains better.
Practical angle:
- In the Hunt Cup, avoid middle draws unless the pace map shows a strong cluster there.
- Look for horses drawn near the pace, not just the “good” side.
Round course (1m2f, 1m4f, 2m+)
Takeaway:
Draw bias is minimal on the round course, but pace bias is huge.
Why:
- The long sweeping bend equalises draw.
- Ascot’s uphill finish punishes front‑runners.
- Hold‑up horses often come home strongly.
Practical angle:
- Favour strong travellers and closers.
- Draw only matters in very large fields where low draws can get boxed in early.
The one non‑obvious insight punters often miss
Ascot’s bias is not fixed — it’s reactive.
The first two straight‑course races of the day often tell you more than the last 10 years of data.
If the first sprint shows:
- One side pulling 3–4 lengths clear → that’s the bias.
- Both sides finishing together → no meaningful bias that day.
If you want, I can also give you:
- A cheat sheet for Royal Ascot week (where bias tends to shift day by day)
- A pace‑map‑based betting strategy for Ascot
- Historical draw‑win percentages by stall number
Which one would help you most?
simon boardman
Mare
Sean Boyce's take on Ascot Draw Bias
Chesham
Sire
Perhaps AI. could improve on the Ability and Consistency Ratings. Given the correct suggestion for formula or Algorithm. It might just be possible. Especially if given the best data . Also sectional data and stride metrics for Flat and AW. With regards NH Racing the Jumping metrics and there are quite a few of those to consider and throw into the mix
Just a thought
Just a thought