Видео

This will not work, I’m working on this you need a lot of steps before just querying you have to break down the document.

The idea is good using RAG, nevertheless the issue is the documents, cause stocks forecast analysis not necessarily will behave based on papers.

I'm guessing Eastern Seaboard cuz man you are up early😮

This is just a very simple python algorithm that automates a chart image to an LLM. What's the point? You can simply upload a chart image yourself, that would be quicker. This is a pointless use of AI and has zero value to a trader.

Very interesting thanks!

For us new followers, "This code requires the 0.2.40 version of yfinance: "pip install yfinance==0.2.40"

Hi, thank you again ❤😊. But why don’t you use other api instead of using chatGPT, it’s not better even with charged version

Hi, how do I run this, it won't let me run the .py file due to an error, do I have to put it in a browser and if so how? Very cool concept btw! It doesnt look messy at all for the amount of code there is!

An excellent video on using AI Agents for stock analysis. Thank you.

Good video. But I have a doubt, now when we can directly upload any pdf or documents inside chatgpt and google ai studio, and get the same results, then why to create these RAG based llm wrappers. What's the point of these?

What is the ta module?

Why not make free agents in Microsoft Copilot Studio?

Thanks again

Want to support the channel? Subscribe to the free Deep Charts Newsletter for more deep learnings: deepcharts.substack.com/

What’s the requirements.txt file?

What a clearly explanation about this topic with FASTHTML. Hopefully more tutor with this framework and Picocss. Many thanks.

🙂

When I try the "run the ai" it is taking forever loading with the message "Analyzing the chart, please wait..." Did anyone experience the same issue, and perhaps figured out to solve?

Cool mate, Learn Alot form this, thankyou

I found this useful reviewed thank you for the information brother

How much does the API cost?

You lost me when you suddenly moved to a jupyter notebook. I had everything installed. I'm lost on the jupyter notebook part.

Can this agentic framework have multiple coding agents working on the same code at the same time?

How was your experience in having the agent produce consistent and incremental improvement?

Want to support the channel? Subscribe to the new Deep Charts Newsletter for more deep learnings: deepcharts.substack.com/

Thank you! How about data privacy, would using locally run private lllm instead of Gemini help?

For someone reason when I try to look at the app via my phone or anywhere outside of the streamlit clouding tells me I don't have access to it.

Thank you so much! It's really helpful!

I've just checked on Google Trends and popularity of "Blue Sky which is a fast growing alternative to Twitter or X" dropped over 75% since 14th November 2024. How is it fast growing? I don't need AI sentiment monitoring to see it is quite opposite....

Top of top 👍👍👍

Thank and many thanks

Awesome!! Keep it Up!

brilliant! thanks!

Excellent video!

Want to support the channel? Subscribe to the new Deep Charts Newsletter for more deep learnings: deepcharts.substack.com/

You will have to go with sequence train and test only. Random will never give you result it gave the last data to model of different date and test over different date that wouldn't work.

I think without model training and RL technique it wouldn't be accurate

Boss how about if we import this llama 3.2 on local host then train over backdated data, in my understanding it is open source , fine tunned and powerfull model with no expense. Will that work from model point of view?

Yfinance can only provide open,close,high,low and vol. What if we have to get all the data like order book, open interest and other. Like for different - different indicators we need different elements.. any guess how we will be getting that information. One more thing about - chart pattern like tradingview has script which will create the chart patterns live and provide the target and stop loss. Can we create anything like that too! I.tried but nothing working out!

Thank you so much for making this video, it is simple and to the point! Is it possible to deploy this app on streamlit cloud services? I am asking because you have stored the data and installed the Ollama model on a local machine.

Some methods are changed in current version eg geo to h3, after change It's worked and thanks for this great video❤❤❤

thank you too much , keep going.❤❤

why not use dspy?

super fire! taking notes. will let you know if i take it to prod

Thank you!

This was a well made video. I appreciate that you kept it simple for people who aren't familiar with the ML/DL lingo. A question came to mind while watching this video though. Have you compared results between Llama 3.2 Vision and something like Gemini 2.0 Flash Thinking? I would be interested to see if the vision capability is a benefit for the LM/VM. To say it another way, these large models are effectively "imagining" everything by language OR they are "seeing" an image by reading the pixel information of the image. I realize that I stepped on a ton of land mines there, but I want to stay focused on the effective results of the models. Would a vision model beat a language model give something like this? I wonder. I know for a fact that Claude Sonnet, Gemini 2.0, and Llama 405b can. Looks like I have a new experiment to conduct. Cheers!

superb!!!!!

Excelent Explanation!! Thanks for sharing your knowledge with us. Have a great 2025!! Thanks again.

Your channel has been one of the best sources of ideas for me in recent times. I appreciate the time you’ve invested. Happy holidays and I can't wait for next year's stuff!

Want to support the channel? Subscribe to the new Deep Charts Newsletter for more deep learnings: deepcharts.substack.com/