Introducing FinAgents

As an educational endeavor, with the launch of Claude 3.7 and while waiting for $NVDA (NVIDIA Corporation) to publish results, I decided to fool around with langchain and created a multi-agent system that simulates a debate between famous investors analyzing stock investment opportunities.

You can find the github repo if you want to check it out.

Currently it is hardcoded to have Warren Buffett, Ray Dalio, Cathie Wood, Peter Lynch and Michael Burry debate about $MSFT (Microsoft) based on analysis provided to them by a team of analyst agents.

github.com/weirdapps/finagents

## Overview

FinAgents is an educational project. It creates a virtual investment committee where AI agents with personas of famous investors like Warren Buffett debate whether stocks are good investments. The system:

1. Gathers real financial data using yfinance 2. Has specialized analyst agents evaluate different aspects of the stock 3. Has investor agents with unique investment philosophies debate the merits 4. Synthesizes a comprehensive investment recommendation

## Key Features

- *Parallel Processing*: Agents run concurrently for faster analysis - *Diverse Perspectives*: 5 famous investor personas with unique philosophies - *Specialized Analysis*: 5 types of analyst agents focus on different aspects - *Real Market Data*: Uses yfinance to pull actual stock information - *Comprehensive Output*: Generates detailed reports saved as Markdown files

The system will analyze Microsoft (MSFT) by default, using all agents in parallel, and save detailed reports to the `results` directory.

## Output Files

For each analyzed stock (e.g., MSFT):

- results/MSFT_decision.md: Final synthesized investment recommendation - results/MSFT_detailed_analysis.md: Detailed reports from all analysts and investors

## Requirements

- Python 3.8+ - Anthropic API key (for Claude 3.7) - Internet connection (for financial data retrieval)

## License: MIT

Happy to get your thoughts.