• REGISTER NOW!! Why? Because you can't do much without having been registered!

    At the moment you have limited access to view all discussions - and most importantly, you haven't joined our community. What are you waiting for? Registration is fast, simple and absolutely free so please, join Join Talking Horses here!

Trainers In Form

walsworth

Journeyman
Joined
Jul 19, 2006
Messages
2,212
Location
North Herts
This is my list of trainers worth looking at for the next couple of days.
Not all of them will have runners today and I don't bet indiscriminately.

Maybe some will find it useful, be lucky whatever!

Appleby, Charlie
Burrows, Owen
Donoghue, Ian
Haggas, W J
Morrison, H
Owen, J P
Tickle, J
 
This is my list of trainers worth looking at for the next couple of days.
Not all of them will have runners today and I don't bet indiscriminately.

Maybe some will find it useful, be lucky whatever!

Appleby, Charlie
Burrows, Owen
Donoghue, Ian
Haggas, W J
Morrison, H
Owen, J P
Tickle, J
Jack Channon, has hit a bit of a purple patch.
Group 3 at Leopardstown last week. Listed success in France Sunday.
Valuable sprint at Windsor last night.
 
I can't code (I chose Latin O Level over Computer Studies - a dead language over the future - in 1977, how bad a choice was that?) but my best friend can and even he uses AI to do the coding for him when I ask for something nowadays - I ask at dawn and have the model minutes after he's looked at it and seconds after he's asked AI to look at it.
 
I can't code (I chose Latin O Level over Computer Studies - a dead language over the future - in 1977, how bad a choice was that?) but my best friend can and even he uses AI to do the coding for him when I ask for something nowadays - I ask at dawn and have the model minutes after he's looked at it and seconds after he's asked AI to look at it.
Now, if we could just find the right shit to put into it.
 
Have you tried modelling your data using AI?
Short answer, yes a bit.

Long answer, in another place where I really post, it has a lot of guys into the tech data analysis side like me, and some are getting heavily into playing around with AI e.g. chatgpt and claude. In one of by invite only sections there was a good back and forth about what we found out.
If you let AI's utilise what's on t'internet you basically get a glorified packaged and formatted google search, which can be really useful for some things but so much for others.

If you don't have the skills, AI can write code but from what I've been told by guys who use it like that you have to check what it's done and it can be something of an iterative process to get it's output correct, or use it as a starting point to finish yourself. But at the moment it's at junior programmer level for what it can do.

The best approach is to feed your own data into an AI and limit it's analysis to that data, but you hit a point where you have to pay due to size of the data or amount of analysis time.

I had a good play about with claude feeding in some of my own bespoke data, which AI couldn't generate, and even I paid so I had some decent bandwidth to use.
I'm used to looking at pages of numbers and/or charts and understanding what it's telling me, and/or first refining it myself to get a clearer view, so I tried feeding in data I'd already analysed to see if the AI told me anything different.
It didn't come out with anything outstandingly new but it helped coalesce my own thinking or cause me to look again at something I'd seen but hadn't given enough thought to. My view is that it works well as an extra pair of eyes but I'd never blindly trust what it tells me
But I need to delve further as that was just from my initial dabbling.

What I'd like to do is repeat the experiment but with a really big dataset but that would mean spending more dosh than I'm willing to spend on AI experimentation at the moment.

AI is not magic , it's just another computer program and like any other computer program, it's a case of nonsense in , nonsense out when feeding data in or the type of questions you ask.

You have to be very precise with the questions you fire at AI or you can get slightly different answers when asking the same thing more than once, also different AI models won't all give the same answers.
 
AI is not magic , it's just another computer program and like any other computer program, it's a case of nonsense in , nonsense out when feeding data in or the type of questions you ask.

You have to be very precise with the questions you fire at AI or you can get slightly different answers when asking the same thing more than once, also different AI models won't all give the same answers.

I agree with that. It’s about much more than just college kids cheating on papers, but it does require nuance, like most other things in life.

I’d push back a bit on the “junior programmer” line. That might be true for some things, but in data analysis and pattern spotting it can already beat a lot of human analysts if you set it up right. The real bottleneck is usually the person using it, not the AI’s ability.
 
I agree with that. It’s about much more than just college kids cheating on papers, but it does require nuance, like most other things in life.

I’d push back a bit on the “junior programmer” line. That might be true for some things, but in data analysis and pattern spotting it can already beat a lot of human analysts if you set it up right. The real bottleneck is usually the person using it, not the AI’s ability.
I stick by my view re writing new code for you to take away and use yourself.
But data analysis and pattern matching is a different realm and not a junior programmer job. There it has shown it can do good stuff if thrown enough cpu at it
 
These are the races that I am considering a bet in, depending on further consideration.

RaceHorseJockeyOdds
3.12 BeverleyDinamo (IRE)Jamin, Pierre-Louis6
6.00 WindsorGloriously SassyLee, Clifford6.5
4.20 BeverleyChillhi (IRE)Hart, Jason2.38
4.35 SalisburyGhaiyyaFanning, Joe5.5
7.05 TramoreDerryville (IRE)Enright, P T3.5
7.15 Chelmsford CityMedinilla (IRE)Dawes, Rose7.5
7.22 CorkChorusHayes, C D4.5
7.30 WindsorAmmes (IRE)Watson, Jason2.63
7.45 Chelmsford CityCrack Shot (IRE)Spencer, Jamie3.75
8.25 CorkThegooseiscooked (IRE)Keane, C T2.1
 
I have a jockey report the same as the trainer one I showed an excerpt of.

I've dabbled with trainer jockey combos before but to do that report for those would mean some new code
 
There should be a permanent pinned thread called: "Ask pawras" here so requests for statistical data searches can be made.

He might tell you to F off and do your own work - I know I would - but if you don't ask you can't get!
 
There should be a permanent pinned thread called: "Ask pawras" here so requests for statistical data searches can be made.

He might tell you to F off and do your own work - I know I would - but if you don't ask you can't get!
I have a thread where I've posted stuff.

Response will be on a case by case basis , in some instances it will be 1.01 to be fk off and when you get there fk off again a bit further
 
Short answer, yes a bit.

Long answer, in another place where I really post, it has a lot of guys into the tech data analysis side like me, and some are getting heavily into playing around with AI e.g. chatgpt and claude. In one of by invite only sections there was a good back and forth about what we found out.
If you let AI's utilise what's on t'internet you basically get a glorified packaged and formatted google search, which can be really useful for some things but so much for others.

If you don't have the skills, AI can write code but from what I've been told by guys who use it like that you have to check what it's done and it can be something of an iterative process to get it's output correct, or use it as a starting point to finish yourself. But at the moment it's at junior programmer level for what it can do.

The best approach is to feed your own data into an AI and limit it's analysis to that data, but you hit a point where you have to pay due to size of the data or amount of analysis time.

I had a good play about with claude feeding in some of my own bespoke data, which AI couldn't generate, and even I paid so I had some decent bandwidth to use.
I'm used to looking at pages of numbers and/or charts and understanding what it's telling me, and/or first refining it myself to get a clearer view, so I tried feeding in data I'd already analysed to see if the AI told me anything different.
It didn't come out with anything outstandingly new but it helped coalesce my own thinking or cause me to look again at something I'd seen but hadn't given enough thought to. My view is that it works well as an extra pair of eyes but I'd never blindly trust what it tells me
But I need to delve further as that was just from my initial dabbling.

What I'd like to do is repeat the experiment but with a really big dataset but that would mean spending more dosh than I'm willing to spend on AI experimentation at the moment.

AI is not magic , it's just another computer program and like any other computer program, it's a case of nonsense in , nonsense out when feeding data in or the type of questions you ask.

You have to be very precise with the questions you fire at AI or you can get slightly different answers when asking the same thing more than once, also different AI models won't all give the same answers.

Sometimes its hard to believe we're friends.

It's a bit like that my name is maximus deridius my name is Jeff thing that goes around.
 

Recent Blog Posts

Back
Top