🚀 IndiaQuant MCP – AI-Powered Market Intelligence System

Production-ready AI system for real-time stock analysis, trading signals, and portfolio simulation using FastAPI and MCP architecture.

⚡ Built with:

• FastAPI backend
• Real-time APIs (yfinance, NewsAPI)
• Options analytics + Black-Scholes Greeks
• Portfolio simulation (SQLite)
• AI-agent compatible MCP tools

👉 Designed as a modular system for real-world financial intelligence applications

IndiaQuant MCP is a real-time AI-powered market intelligence system built using the Model Context Protocol (MCP).
It provides live stock data, trading signals, options analytics, sentiment analysis, and portfolio simulation using 100% free APIs.

The system exposes these capabilities as MCP-compatible tools so an AI agent (like Claude Desktop) can query and analyze financial markets in real time.

---

Project Architecture

AI Agent (Claude / AI Assistant)
        │
        ▼
MCP Tool Server (FastAPI)
        │
        ├── Market Data Engine
        ├── Signal Generator
        ├── Options Analyzer
        ├── Greeks Calculator
        ├── Portfolio Manager
        ├── Sentiment Analyzer
        ├── Market Scanner
        └── Sector Heatmap
        │
        ▼
External APIs
   ├── Yahoo Finance (yfinance)
   ├── NewsAPI
   └── Alpha Vantage (optional)

The system is designed as a modular financial intelligence platform, where each component provides a specific capability.

---

Project Structure

indiaquant-mcp
│
├── app
│   ├── market_data
│   ├── signals
│   ├── options
│   ├── analytics
│   ├── portfolio
│   ├── decision          # decision layer v1 (normalize → fuse → validate)
│   └── mcp
│       ├── mcp_server.py    # FastAPI + OpenAPI
│       └── stdio_server.py  # native MCP stdio (Claude Desktop / Cursor)
│
├── docs
│   └── decision_layer_first_draft.md
│
├── tests
│   └── test_decision_engine.py
│
├── .github/workflows
│   └── ci.yml
│
├── screenshots
│   ├── live_price.png
│   ├── signal.png
│   ├── trade.png
│   └── heatmap.png
│
├── main.py
├── pytest.ini
├── CHANGELOG.md
├── requirements.txt
└── README.md

---

MCP Tools Implemented

The following MCP / HTTP tools are implemented.

Tool	Description
get_live_price	Fetches live stock price and market data
generate_signal	Generates BUY/SELL/HOLD signal using technical indicators
get_options_chain	Retrieves options chain data
calculate_greeks	Computes Black-Scholes Greeks
place_virtual_trade	Simulates buy/sell trades
get_portfolio_pnl	Calculates portfolio profit and loss
analyze_sentiment	Performs sentiment analysis on financial news (NEWSAPI_KEY env)
detect_unusual_activity	Detects unusual options activity
scan_market	Scans market for oversold stocks
get_sector_heatmap	Displays sector performance heatmap
fuse_market_decision	Decision layer v1: fuses technical + sentiment + options into unified direction, edge score, and validation
fuse_decision_manual	Same fusion engine with caller-supplied normalized signals (tests / custom pipelines)
schemas/decision_layer (GET)	JSON Schema bundle for decision-layer Pydantic models (integrators / contract tests)

All tools return live market data using free APIs where applicable. See docs/decision_layer_first_draft.md for schema, examples, and fusion rules; CHANGELOG.md summarizes decision-layer v1. Run tests: pytest (see pytest.ini). CI: .github/workflows/ci.yml.

---

Core Modules

Market Data Engine

Uses yfinance to fetch real-time market data.

Capabilities:

• Live stock prices
• Historical OHLC data
• Volume and price change analysis
• Supports NSE and global stocks

Example response:

{
"symbol": "RELIANCE",
"price": 1418.6,
"change_percent": 3.17,
"volume": 34897
}

---

AI Trade Signal Generator

Generates trading signals using technical indicators:

Indicators used:

• RSI
• MACD
• Bollinger Bands

Signal output: BUY / SELL / HOLD confidence score

Example:

{
"symbol": "RELIANCE",
"signal": "BUY",
"confidence": 40
}

---

Options Chain Analyzer

Retrieves options data and performs analysis including:

• Open Interest tracking
• Max Pain calculation
• Options volume comparison
• Unusual activity detection

This helps identify potential institutional trading behavior.

---

Greeks Calculator

Implements the Black-Scholes model from scratch.

Greeks calculated:

• Delta
• Gamma
• Theta
• Vega

Example:

{
  "delta": 0.2265,
  "gamma": 0.026248,
  "theta": -0.06355,
  "vega": 0.172592
}

---

Portfolio Risk Manager

Simulates a virtual trading portfolio using SQLite.

Features:

• Place virtual buy/sell trades
• Track portfolio positions
• Real-time PnL calculation
• Trade history storage

Example: POST /place_virtual_trade

{
"symbol": "RELIANCE",
"qty": 1,
"side": "BUY"
}

---

Sentiment Analysis

Uses NewsAPI to analyze market sentiment from financial news.

Process:

1. Fetch recent headlines
2. Score sentiment based on keywords
3. Generate sentiment signal

Example output:

{
  "symbol": "RELIANCE",
  "sentiment_score": 2,
  "signal": "POSITIVE"
}

---

Market Scanner

Scans multiple stocks to find oversold opportunities.

Criteria: RSI < 30

Example response:

[
{
"symbol": "AAPL",
"RSI": 28.3,
"signal": "OVERSOLD"
}
]

---

Sector Heatmap

Analyzes sector performance by aggregating stock movements.

Example output:

[
{"sector": "IT", "change_percent": 0.35},
{"sector": "BANKING", "change_percent": -2.82},
{"sector": "ENERGY", "change_percent": -0.78},
{"sector": "AUTO", "change_percent": -4.6}
]

---

Technologies Used

Core stack:

• Python
• FastAPI
• SQLite
• yfinance
• pandas
• numpy
• NewsAPI

Libraries: fastapi uvicorn pandas numpy yfinance newsapi-python sqlite3

---

Installation

Clone repository

git clone https://github.com/sowjanya5751/indiaquant-mcp.git

cd indiaquant-mcp

Create virtual environment

Windows

python -m venv venv venv\Scripts\activate

Linux / Mac

python -m venv venv source venv/bin/activate

Install dependencies

pip install -r requirements.txt

---

Running the MCP Server

Option A — FastAPI (HTTP tools + OpenAPI)

Start the server:

cd indiaquant-mcp
pip install -r requirements.txt
uvicorn app.mcp.mcp_server:app --reload

Server will start at:

http://127.0.0.1:8000

Option B — Native MCP (stdio, Claude Desktop / Cursor)

The repo also exposes an official MCP server over stdio using the Python mcp SDK (FastMCP), including fuse_market_decision and core market tools.

From the repo root:

PYTHONPATH=. python -m app.mcp.stdio_server

Example Claude Desktop (claude_desktop_config.json) fragment:

{
  "mcpServers": {
    "indiaquant": {
      "command": "python3",
      "args": ["-m", "app.mcp.stdio_server"],
      "cwd": "/absolute/path/to/indiaquant-mcp",
      "env": {
        "PYTHONPATH": ".",
        "NEWSAPI_KEY": "your-key-optional"
      }
    }
  }
}

---

API Endpoints

Endpoint	Method	Description
/get_live_price	POST	Fetch live stock price
/generate_signal	POST	Generate trading signal
/get_options_chain	POST	Retrieve options data
/calculate_greeks	POST	Compute Black-Scholes Greeks
/place_virtual_trade	POST	Execute simulated trade
/get_portfolio_pnl	GET	Calculate portfolio PnL
/analyze_sentiment	POST	Analyze financial news sentiment
/fuse_market_decision	POST	Decision layer v1: fused direction + edge + validation
/fuse_decision_manual	POST	Fuse caller-supplied normalized signals
/schemas/decision_layer	GET	JSON Schema bundle for decision models
/detect_unusual_activity	POST	Detect unusual options activity
/scan_market	GET	Find oversold stocks
/get_sector_heatmap	GET	Sector performance overview

---

API Documentation

Interactive API documentation is available at:

http://127.0.0.1:8000/docs

Swagger UI allows testing all MCP tools directly.

---

Design Decisions

FastAPI was chosen because:

• High performance async framework
• Automatic API documentation
• Ideal for MCP tool integration

SQLite was used because:

• Lightweight database
• Perfect for portfolio simulation
• Easy local deployment

yfinance provides:

• Free stock market data
• Historical price access
• Options chain support

---

Future Improvements

Possible extensions:

• Real-time WebSocket streaming
• Machine learning trading models
• Redis caching for faster data retrieval
• Cloud deployment (AWS / GCP)
• Advanced portfolio risk analytics

---

Assignment Requirements Fulfilled

✔ Real-time market data
✔ 10 MCP tools implemented
✔ Options analysis and Greeks calculation
✔ Sentiment analysis using NewsAPI
✔ Virtual trading portfolio
✔ Modular system architecture
✔ API-based MCP server compatible with AI agents

---

API Demo

Live Price

Trade Signal

Portfolio Trade

Sector Heatmap

Conclusion

IndiaQuant MCP demonstrates how AI agents can interact with financial markets through modular tools and real-time data pipelines.

The system combines quantitative analysis, market intelligence, and AI integration into a unified platform capable of supporting advanced trading insights.