Anybody using AI models with their trading?

Tools like Chatgpt, Gemini, Deepseek, Claude and LLama i think are game changers for us.

If you don’t understand a topic or just want more details, it’s like a google search away to get your questions answered with one of the tools.

But it’s better because you can ask followup questions that remember what you already asked and what information the AI model already gave you.

It can even help you code indicators for your favorite trading platform. I tried to convert a MT4 plugin to Pinescript for Tradingview. I didn’t getting it working completely, but in just a couple hours, I got the bones of the indicator working. It’s pretty crazy really powerful.

Have you guys used it for anything?

I’ve used it to put my strategy into words, and not just some random collection of things I do or follow that makes me take a trade. That part for me is like the coolest part. It’s talking to the model, telling it how I trade, and it takes what I say and spits out the “rules” of my strategy.

It can even create a strategy for you if you tell it what you trade, what session you trade, timeframe, currency pair, all that, and give you tips on how to approach entries, exists, risk management, SLs, TPs, all of that. I mean it regurgitates all the stuff I learned in the School and from other traders here, but it’s crazy how thorough it is.

And it’s really good at summarize topics into easy to understand explanations. You can literally tell it to explain a topic to you like you’re a kid.

Anybody else using AI with their trading?

5 Likes

These AI models use “information already online” to learn on, don’t they?

So if it’s a subject about which there’s actually more bad information than good information already online (such as maybe trading???), does that mean the answers you’ll get to your questions are more likely to be bad ones than good ones?

I do agree, from the things I’ve seen quoted in forums from AI sources, that their use of language isn’t unimpressive at all. I wonder, though, whether that appearance might lend them some artificial credibility they don’t necessarily deserve?

I have no personal experience of these things, and am asking - genuinely - out of interest, not to try to be dismissive of them!

1 Like

Hi @samewise,

I don’t really trust AI, as I had mentioned many time, I have been working with AI, specifically in artificial neural network. Creating math model, build the neural node then do training to make the model smart is my daily side jobs :sweat_smile:

For ChatGPT and Gemini, I think them as assistance to do some expert casually do. For example you want an opinion about something, you want to know what in general, ChatGPT and Gemini will be perfect for it. You can ask, how to code something from EA to PineScript, the AI tools can help you. Because, they have data to summarize solution to solve your problems. But when you need some breakthrough, they are dull for it.

ANother example, I ask about how to trade XAUUSD, their will give answer base on its data pool. When the pool is all dump people, their answer is also dumb. Remember most trader, 90%, loss in market. They become source of data for ChatGPT and Gemini, then when we ask questions, AI tools actually provide answer from these 90% unsuccessful trader. :sweat_smile:

Unless ChatGPT and Gemini have data pool from well known practical traders, I then believe they will give outstanding answers. But, do you thing these 10% superb traders willing to share their secrets?

So, I will use AI for lazy stuffs, just like … I want to know how how EMA can be used. AI will show you a good long list. You may also can ask AI, how to trade XAUUSD. Then you can take the answer as inspirations to create something new.

I just ask GPT, what is the best methodology to trade XAUUSD. The answer is MA Crossover. When I put it in EA … it’s a total garbage :sweat_smile:

2 Likes

I’m not using it but yes tools like Chatgpt have made things easier. Have you found any tool which works best for your trading setup?

Nope that’s a really good point. I think they call it garbage in garbage out. But they try to help that problem, so LLMs can filter stuff, they can be pretrained, and they can do fact checking.

“But who checks the fact-checkers?!?!?” Yea, I don’t know.

I think with math, and statistics and patterns, there’s less gray maybe.

Yea, I get that completely.

Yea, they’re great researchers, summarizing stuff. Maybe connecting two or three points.

On the indicator building, I think I just scratched the surface. For me the thing was crazy because I don’t have a coding background whatsoever. So it was cool to see something work with a little bit of trying.

What kind of breakthroughs are you looking for? Is this for trading?

No you’re so right. I don’t think people understand that part. Garbage in garbage out.

That’s a big fat NO! :sob:

That’s what I’m doing, just like this.

3 Likes

Not really the setups themself. It’s more like trading concepts and what things to look for. Like with MA’s or MACD or using different time frame analysis to find a trade entry, it’s trying to connect the dots of all the stuff I have on my screen.

So price action, indicators (volume, MACD, market structure, FVGs, SMC), trendlines and figuring out what they’re telling me.

I use Chatgpt, Gemini and Deepseek right now, just to see what they say to the same question. Really interesting.

2 Likes

You’re either an “enthusiast” or an “addict” - your choice! :grinning:

Yes. The Wall Street Journal now includes some article about AI almost every day! I wonder whether maybe Babypips forum is thinking of opening a new sub-forum here, under “Tech” but specifically for AI-related subjects? Would that be a good idea?

Ha, I just want to make my life easier. And learn more things!

You mean specific to trading?

1 Like

Yes - sorry - I meant “trading-related”!

1 Like

Hi @samewise,

Yes my work in AI, mostly related to trading in this case. I have tried many math model to find out new repetitive pattern in market. We need to provide the math model for each case. We cant train AI to learn to create math model.

Example Math model for Elliot Wave:

EW = W22xW1 and W4<W3 and W5<W3

Above math model only available for common EW type. There are many variation EW for every instrument. To have all of the EW model, we need to apply those models into our engine. When we are not careful, it become garbage :smiley:

As human, we have capability to learn to build something new by our creativity, create something from nothing. AI “creates” something rely on data we feed.

I don’t say AI is useless, AI has its practical benefit on certain degree. :slightly_smiling_face:

1 Like

ChatGPT sucks at developing strategies. I challeneged it to a competition of who could make the most money on EURUSD with an algorithm and it had about 10 different attempts that all lost a lot.

What it was better at was refining my algorithm to improve the signal and make the win rate go much higher. The down side was that it only traded about twice a month and I lost interest with pursuing that strategy.

Which trading-focused GPT do you use, out of interest?

Do you think AI tools give better insights than others?

Couldn’t hurt. You could start the first thread!

I guess this is good. If AI gets that’s smart, what’s the point of humans. Sounds like Terminator the movie and… what’s the name… Skynet!

1 Like

Oh that’s crazy! I guess it can access historical data, or did you somehow feed it live data, or what?

That’s cool though. I wonder if it could learn from the attempts and then make changes to it’s strategies??

It has access to data that it can analyse itself. But in this case, it just wrote code.

When it made a strategy after analysing data, it was better but not great.

Yea I think it will only use strategies that already exist and not actually create anything new that’s like groundbreaking.

Which makes sense. It’s not really “thinking”.

1 Like