IndiaQuant MCP Server

A Model Context Protocol (MCP) server for Indian stock market analysis providing real-time market data, technical analysis, options analytics, and portfolio management using 100% free APIs.

Overview

IndiaQuant MCP provides 11 tools for NSE/BSE market analysis:

• Live market data (prices, volume, percentage changes)

• Technical analysis (RSI, MACD, Bollinger Bands)

• Options analytics (Black-Scholes Greeks calculated from scratch)

• Sentiment analysis (news-based scoring with NewsAPI + TextBlob)

• Market scanning (NIFTY 50 overbought/oversold detection)

• Portfolio management (virtual trading with P&L tracking)

• Sector analysis (IT, Banking, Auto, Pharma, FMCG heatmaps)

Architecture

Module Structure

IndiaQuant/
├── server.py                    # MCP server entry point (stdio transport)
├── market_data/
│   └── fetcher.py              # Live NSE/BSE price fetching (yfinance)
├── trading_signals/
│   └── signal_generator.py     # RSI, MACD, Bollinger Bands (pandas-ta)
├── options_analysis/
│   ├── options_analyzer.py     # Options chain data (yfinance)
│   ├── greeks.py               # Black-Scholes Greeks implementation
│   └── unusual_activity.py     # Volume/OI spike detection
├── utils/
│   ├── sentiment.py            # News sentiment analysis
│   └── market_scanner.py       # NIFTY 50 scanner, sector heatmap
├── portfolio_management/
│   └── portfolio_manager.py    # Virtual portfolio with SQLite
└── database/
    └── portfolio.db            # SQLite database

Design Decisions

1. Modular Architecture

• Each module handles a specific domain (market data, signals, options, etc.)

• Clean separation of concerns for maintainability

• Easy to extend with new tools

2. 100% Free API Stack

• yfinance: Live NSE/BSE prices, historical data (unlimited, free)

• NewsAPI: News headlines for sentiment (100 requests/day free tier)

• pandas-ta: Technical indicators (open source)

• TextBlob: Sentiment scoring (open source)

• Custom Black-Scholes: Greeks calculation from scratch

3. MCP Protocol Implementation

• Stdio transport for maximum compatibility

• Proper tool registration with JSON schemas

• Type-safe parameter validation

• Structured error handling

Trade-offs & Limitations

What Works:

• Live NSE/BSE stock prices for all securities

• Technical indicators (RSI, MACD, Bollinger Bands)

• Greeks calculation for any stock option

• News sentiment analysis

• Portfolio tracking with real-time P&L

• Sector performance analysis

Known Limitations:

• NSE Options Chain Data: yfinance has limited support for NSE options chains. Works reliably for US stocks but inconsistent for Indian stocks due to Yahoo Finance's data coverage. For production, would use NSE Official API or broker APIs (Zerodha, Angel One).

• NewsAPI Rate Limits: Free tier allows 100 requests/day

• Index Options: NIFTY/BANKNIFTY index options not supported via yfinance

Design Justification:

• Chose yfinance for zero-cost, zero-setup advantage despite NSE options limitations

• Prioritized breadth of features (11 tools) over depth

• Greeks calculation works perfectly when stock price is provided

• Assignment requires 100% free APIs - no paid alternatives allowed

MCP Tools (11 Total)

1. health_check

Check if server is running.

Parameters: None

Returns: {"status": "server running"}

2. get_live_price

Fetch live NSE stock price.

Parameters:

• symbol (string): Stock symbol (e.g., 'RELIANCE', 'TCS')

Example Response:

{
  "symbol": "RELIANCE.NS",
  "current_price": 1408.40,
  "change_percent": 1.54,
  "volume": 5234567
}

3. generate_trading_signal

Generate BUY/SELL/HOLD signals using RSI, MACD, Bollinger Bands.

Parameters:

• symbol (string): Stock symbol

• period (string, optional): '1d', '5d', '1mo', '3mo', '1y' (default: '3mo')

Example Response:

{
  "symbol": "TCS",
  "signal": "BUY",
  "confidence": 75,
  "reasoning": "RSI oversold (28.5), MACD bullish crossover"
}

4. get_options_chain

Fetch options chain with calls, puts, OI, PCR.

Parameters:

• symbol (string): Stock symbol

• expiry (string, optional): Expiry date YYYY-MM-DD

Note: Limited data availability for NSE stocks (see Limitations section)

5. calculate_greeks

Calculate Black-Scholes Greeks from scratch.

Parameters:

• symbol (string): Stock symbol

• strike_price (number): Strike price

• expiry_date (string): YYYY-MM-DD

• option_type (string): 'call' or 'put'

• volatility (number, optional): IV (default: 0.25)

• risk_free_rate (number, optional): Rate (default: 0.065)

Example Response:

{
  "greeks": {
    "delta": 0.575,
    "gamma": 0.0057,
    "theta": -1.107,
    "vega": 1.075,
    "rho": 0.295
  }
}

Implementation: Pure mathematical Black-Scholes using scipy.stats

6. detect_unusual_activity

Detect volume/OI spikes in options.

Parameters:

• symbol (string): Stock symbol

• volume_threshold (number, optional): Spike multiplier (default: 3.0)

• oi_threshold (number, optional): OI multiplier (default: 2.0)

7. scan_nifty50

Scan NIFTY 50 for overbought/oversold stocks.

Parameters:

• rsi_threshold (number, optional): RSI threshold (default: 30)

• condition (string, optional): 'oversold' or 'overbought'

8. get_sector_heatmap

Sector performance heatmap.

Parameters:

• period (string, optional): '1d', '5d', '1mo', '3mo' (default: '1d')

9. place_trade

Place virtual trade in portfolio.

Parameters:

• symbol (string): Stock symbol

• quantity (number): Number of shares

• side (string): 'buy' or 'sell'

• price (number, optional): Price (defaults to live market price)

10. get_portfolio_pnl

View portfolio positions with real-time P&L.

Parameters: None

11. analyze_sentiment

Analyze news sentiment using NewsAPI + TextBlob.

Parameters:

• symbol (string): Stock symbol

• days (number, optional): Days to look back (default: 7)

Note: Free tier limited to 100 requests/day

Installation

Prerequisites

• Python 3.8+ (tested on 3.13.5)

• Claude Desktop or any MCP-compatible client

• NewsAPI Key (free: https://newsapi.org/)

Setup

# Clone repository
git clone https://github.com/Shirshak-dugtal/mcp.git
cd mcp

# Install dependencies
pip install -r requirements.txt

# Configure environment
echo "NEWS_API_KEY=your_key_here" > .env

# Test server
python server.py

Dependencies

mcp>=1.0.0
yfinance>=0.2.40
pandas>=2.0.0
pandas-ta>=0.4.0
numpy>=1.24.0
scipy>=1.10.0
newsapi-python>=0.2.7
textblob>=0.18.0
python-dotenv>=1.0.0

Running the Server

Connect to Claude Desktop

Windows:

Edit %APPDATA%\Claude\claude_desktop_config.json:

{
  "mcpServers": {
    "IndiaQuant": {
      "command": "python",
      "args": ["D:/code/mcp/IndiaQuant/server.py"]
    }
  }
}

Replace path with your actual installation path.

macOS/Linux:

Edit ~/.config/Claude/claude_desktop_config.json with appropriate paths.

Restart Claude Desktop completely (exit from system tray).

Verify: Settings → Developer → Local MCP Server → IndiaQuant should show green checkmark.

Example Tool Responses

get_live_price:

{"symbol": "RELIANCE.NS", "current_price": 1408.40, "change_percent": 1.54}

generate_trading_signal:

{"symbol": "TCS", "signal": "BUY", "confidence": 75, "reasoning": "RSI oversold, MACD bullish"}

calculate_greeks:

{"delta": 0.575, "gamma": 0.0057, "theta": -1.107, "vega": 1.075}

scan_nifty50:

{"matches": [{"symbol": "TATAMOTORS", "rsi": 28.3, "price": 945.20}]}

get_portfolio_pnl:

{"total_invested": 44580, "total_pnl": 314, "pnl_percent": 0.70}

License

MIT License