AI Trading Chatbot

This project implements an AI chatbot to provide personalized trading recommendations using Bayesian networks. The chatbot leads users through the process of constructing a Bayesian network tailored to their specific trading scenario, and utilizes the network to offer data-driven insights.

Overview

• Implements conversational AI using GPT-4 via LangChain and Streamlit
• Allows users to describe their unique trading scenario
• Dynamically generates relevant nodes and edges for a Bayesian network
• Incorporates real-time financial data via FRED and YFinance APIs
• Estimates conditional probability distributions using maximum likelihood
• Performs inferences on the Bayesian network
• Provides trading recommendations based on probabilistic analysis

Features

• Interactive Node Creation: Users can customize the nodes in their Bayesian network, adding, removing or modifying them through conversation.
• Intelligent Edge Recommendations: The chatbot suggests potential edges based on the defined nodes, maintaining network coherency.
• Real-Time Data Integration: Financial time series data is integrated from FRED and YFinance to inform the probability distributions.
• Conditional Probability Estimation: CPDs are estimated using maximum likelihood estimation given the edges and real-time data.
• Trading Recommendations: The chatbot analyzes the Bayesian network to recommend optimal trading decisions based on probabilistic inferences.
• Natural Conversation: The entire interaction from node creation to final recommendations happens through natural dialogue powered by GPT-4.

Installation Requirements:

Python 3.7+ Streamlit LangChain pgmpy pandas yfinance fredapi bash

Copy code

pip install streamlit langchain pgmpy pandas yfinance fredapi

Usage

Copy code

streamlit run risk_bot/🤖_Bot.py

The app will be served at http://localhost:8501. Follow the conversational prompts to construct your Bayesian network and receive trading recommendations tailored to your scenario.

Demo

A video demo of the app can be found here: https://github.com/pareshraut/Automated-Bayesian-Networks/issues/2#issue-2079936495)https://github.com/pareshraut/Automated-Bayesian-Networks/issues/2#issue-2079936495

References The core methodology was adapted from:

LangChain library pgmpy library