📈 FinSight – Multi-Agent Stock Market Intelligence

FinSight is a MERN Stack + Python AI project that brings together real-time financial data, AI-driven analysis, and multi-agent intelligence to empower investors, brokers, and companies with actionable market insights.

---

🏗️ Project Structure

The project is organized into three major folders:

• frontend/ → React.js (component-based UI)
• backend/ → Node.js + Express + Mongoose (MongoDB Atlas as database)
• python-backend/ → FastAPI (Python ML/AI services & endpoints)

---

⚙️ Tech Stack

• Frontend: React.js, Lightweight Charts (TradingView style), Plotly.js
• Backend: Node.js, Express.js, MongoDB Atlas, Clerk (Authentication)
• Python Backend: FastAPI, CrewAI (multi-agent framework), Transformers, ChromaDB
• Cloud & APIs:
  • Angel One SmartAPI (Candlestick data)
  • Yahoo Finance (Balance sheet & ratios)
  • News API (Real-time financial news)
  • Cloudinary (CSV file storage)
• LLM: Gemini 2.5 Pro (for multi-agent reasoning & report generation)

---

🧠 Multi-Agentic AI Pipeline

Built using Crew AI, the pipeline consists of 6 agents, all connected via MCP (Model Context Protocol):

1. 🔎 Web Scraping Agent – extracts market data
2. 📊 Real-Time Candlestick Agent – fetches OHLCV candlestick data
3. 📑 Balance Sheet Agent – pulls company financial statements
4. ⚖️ Ratios Agent – computes financial ratios in real-time
5. 📰 News Agent – scrapes and analyzes financial news
6. ✍️ Summarizer Agent – consolidates all data into human-readable insights

Each agent uses specialized tools and produces structured outputs (CSV), stored in Cloudinary for downstream tasks.

---

🔧 Tools Integrated

• 📈 Candlestick Data Tool – Powered by Angel One Smart Connect SmartAPI
• 🏦 Financials Tool – Balance sheet & ratios from Yahoo Finance
• 📰 News Tool – Real-time financial news with sentiment analysis

---

🌐 Core Features (Frontend)

🔑 Role-Based Access

• Investor
• Broker
• Company
  Authentication is handled using Clerk (JWT & sessions).

📊 Real-Time Candlestick Charts

• Interactive charts (TradingView style) with multiple timeframes: 1m, 15m, 1h, 1d, 1w
• WebSocket updates every minute
• Served via FastAPI endpoints

🏦 Real-Time Financial Data

• Balance sheet & ratios (FastAPI + Yahoo Finance)
• News + Sentiment Analysis (Transformers)
• Visualization via Plotly.js

🤖 Agentic RAG Chatbot

• Knowledge base built from candlestick data, ratios, balance sheets, and summaries
• ChromaDB for vector embeddings
• Context-aware financial Q&A

📑 Agentic Report Generator

• Executes all 6 agents in sequence
• Generates a Markdown report for any company:
  • Market summary
  • Technical indicators (RSI, MACD, EMA, candlestick patterns)
  • Financial ratios & balance sheet analysis
  • News sentiment
  • Recommendation: Buy / Sell / Hold

---

🚀 Future Plans

• 📂 Build a custom dataset combining:

  • OHLCV candlestick data
  • Technical indicators (RSI, MACD, EMA, etc.)
  • Financial ratios
  • News sentiments

• 🤖 Train Deep Learning LSTM models for:

  • Price prediction
  • Time series forecasting
  • Pattern recognition

---

📌 Installation & Setup

1️⃣ Clone the repository

git clone https://github.com/yourusername/finsight.git
cd finsight

2️⃣ Install dependencies

Frontend

cd Frontend
npm i
npm run dev

**Backend (Use your own API secrets) **

cd Backend
npm i
npm run dev

Python Backend

cd Python Backend
py -3.10 -m venv venv
venv\Scripts\activate   # Windows
source venv/bin/activate # Mac/Linux

pip install -r requirements.txt
uvicorn app:app --reload

👨‍💻 Contributor

Manil Modi (@Manil Modi ) – Project Lead & AI Developer