Skip to content

FinMAS - Financial Analysis Multi-agent System#

This app uses LLM agents organized in a multi-agent system to analyze financial data and perform financial tasks. The app is developed during the final capstone project of the WorldQuant MSc in Financial Engineering.

It is meant as a practical and educational app that demonstrates the state-of-the-art of LLM models applied to tasks in the financial domain, and with an extra focus on open source models and packages.

Please visit GitHub repo for further information.

See our Tutorial to get started, and see some example outputs to understand better what the multi-agent systems can do.

Features#

  • Extracting insights from unstructured data: Analyze unstructured data such as SEC filings and News articles together with fundamental company data.
  • Transparent: Get insight into token usage, performance and the data fed to the system to gain confidence in the result.
  • Configurable: Adjust parameters and model selection to optimize the performance of the system.
  • Ticker focused: Select a major ticker listed on NASDAQ or NYSE for analysis.
  • Multiple LLMs supported: Use LLM from an hosted provider such as Groq or HuggingFace, or OpenAI GPT models.
  • Multiple agent systems supported: Use defined crews of agents to perform dedicated analysis on news or SEC filings, or combine them for a final analysis.
  • Query directly the data source: Use the same tool as the agents to query the data source directly from the UI. Powered by llama-index's query engine.

The following screenshots illustrate a output from a combined analysis crew and the main dashboard.

Combined analysis#

Main dashboard#