An AI based stock analyzer using LLM and Langchain📈

[jerry95]

https://preview.redd.it/733v3sk6n0h...bp&s=b4ebcf15c12685c3f3c0b476aef894c11759236e

An interesting application of Language Models and Langchain in the Finance Domain -

Sharing a fun weekend project that I recently completed: the "Stock Analyzer Bot". As an investment enthusiastic person without extensive knowledge in the finance domain, I often end up referring to some finance youtuber's videos or a site on the internet for the fundamental analysis of stocks. To assist in such situations, I developed a stock analyzer bot based on LLM, which gathers up-to-date information about stock such as 1) stock price, 2) Company financials 3) Recent company-related news. The bot then considers all this information to conduct analysis using language models. You can even get positives and negatives about the company's financials, which will certainly help when making an investment decision.

You can ask queries like- "Is it a good time to invest in Yes Bank?" or "How are the current financials of reliance industries looking" and boom within a minute you are presented with a comprehensive financial analysis based on recent data. Of course, It is not recommended to rely fully on the analysis provided by the bot. It seems like a good starting point. And yeah, I agree the possibilities are endless with LLMs.

GIthub- https://github.com/Pranav082001/stock-analyzer-bot

Blog An AI based stock analyzer using LLM and Langchain | by Pranav Kushare | Aug, 2023 | Medium

 

[dothns]

@jerry95 Nice project - question though why use langchain and not open ai api directly? What is the benefit or added value of langchain here

 

[jerry95]

@dothns Valid question: If you see the code or git repo I have experimented with both the approaches

1) First with langchain agent: Which dynamically access the tools and tries to perform action based on the thought. Its a continuous thought and action process. The problem with this is stock analysis being a complex task it often get stuck in the infinite loop

2) Second only with openai API, and predefined prompt. In this I am directly feeding required stock information. This is a simple approach and works great