Skip to content

Latest commit

 

History

History
139 lines (103 loc) · 4.66 KB

README.md

File metadata and controls

139 lines (103 loc) · 4.66 KB

FinBot

Project Description

This project is a multi-agent system designed to function as a Junior Investment Analyst. It assists real analysts in evaluating publicly traded companies by performing both fundamental and technical analyses. The system simulates collaboration between specialized agents to generate detailed financial analysis for user-selected company.

example

Capabilities

  • Fundamental Analysis:
    Calculate and report:

    • Return on Equity (ROE)
    • Return on Assets (ROA)
    • Net Profit Margin
    • Gross Margin
    • Debt to Equity Ratio
    • Interest Coverage Ratio

  • Technical Analysis:
    Generate and visualize:

    • Comparison of closing prices
    • Simple moving averages (given window size or short and long windows)
    • Exponential moving averages (given span)

  • Company Info:

    Provides basic company info such as business summary.

Setup and Running Instructions

Prerequisites

  • Python 3.8+
  • Virtual environment (optional but recommended)

Installation

  1. Clone the repository:

    git clone https://shahd2@bitbucket.org/shah-deep/finbot.git
cd FinBot

  2. Set up a virtual environment:

    python -m venv env  
source env/bin/activate  # For Linux/macOS  
.\env\Scripts\activate   # For Windows

  3. Install dependencies:

    pip install -r requirements.txt

  4. Obtain the Cohere API key:

  5. Add the API key to the environment:

    • Create a .env file in the project directory: 
      COHERE_API_KEY=<your-api-key>

  6. Run the application:

    • Start the backend server: 
      uvicorn backend.server:app --host 127.0.0.1 --port 8000
    • Start the front end: 
      python frontend/app.py
    • Access the Webpage: 
      http://127.0.0.1:8050/

Architecture and Agent Design

System Architecture

  • Front End:

    • Built with Dash and Flask, the front end provides an intuitive interface for user interactions.
    • Communicates with the backend via WebSocket connections for real-time updates.

  • Backend Server:

    • Powered by FastAPI, the backend server listens for WebSocket connections and processes incoming queries.
    • Upon connection, the server initializes the agent system to handle user requests dynamically.

Agent System Architecture

The multi-agent system is designed with the following components:

  • Supervisor Agent:

    • Acts as the central controller.
    • Forwards user messages to the appropriate agent:
      • Fundamental Analysis Agent
      • Technical Analysis Agent
      • Company Info Agent

  • Fundamental Analysis Agent:

    • Handles requests for financial metrics such as ROE, ROA, Net Profit Margin, etc.
    • Retrieves and calculates the necessary data before returning the results.

  • Technical Analysis Agent:

    • Handles requests for price trends and moving averages.
    • Coordinates with:
      • Data Gathering Agent: Fetches required data from financial data sources.
      • Plotting Agent: Generates visualizations such as line graphs and moving average comparisons.

  • Company Info Agent:

    • Provides basic company info such as company name and business summary.

Design Highlights

  • WebSocket Communication: Ensures low-latency, real-time interactions between the front end and backend.
  • Scalable Agent Framework: Allows for easy integration of additional agents in the future.
  • Layered Agent Design: Modularized agents to handle specific tasks, improving clarity and scalability.

Challenges Faced and Solutions

  1. Challenge: Passing large data among the agents.

    • Solution: Stored the data on the server and passed the reference information (e.g., IDs or paths) to retrieve them when required.

  2. Challenge: Communicating between Dash callbacks and WebSocket connections.

    • Solution: Initially tried dash_extensions WebSocket, but it had issues sending messages on startup. Developed a custom connection module using asyncio, enabling WebSocket communication to run in a separate thread seamlessly.

Limitations & Future Work

Technical analysis is performed only on most recent data. To get it for user specified dates, we need to further process the user input to obtain the specific date range.