金融数据分析协议服务器

Tearsheet

Introduction

Tearsheet is a Model Context Protocol (MCP) server designed to enable LLMs (such as Claude, Cursor, or GPT-based agents) to retrieve, analyze, and visualize stock prices and financial report data. It provides a set of robust tools for quantitative trading research, investment analysis, and financial education.

Usage Scenarios

• LLM-driven trading analysis: Let LLMs fetch and analyze stock data, financial statements, and technical indicators to generate trading insights.
• Financial metric calculation: Compute key ratios and metrics from income statements, balance sheets, and cash flow tables.
• Visualization: Generate price charts (line or candlestick), financial metric trends, normalized comparisons, and trading signal charts.
• Backtesting: Run a moving-average crossover strategy and get performance stats (total return, Sharpe ratio, max drawdown, win rate).
• Automated research: Integrate with LLMs to answer questions like "What is the ROE of AAPL?", "Show me the last 6 months of MSFT price data.", or "Plot buy/sell signals for TSLA."

Features

• Retrieve real-time and historical stock prices (single or multiple tickers)
• Retrieve dividends and stock splits history
• Retrieve general ticker info and fundamentals
• Extract and analyze financial statements (income statement, balance sheet, cash flow)
• Calculate key financial metrics (gross margin, operating margin, net profit margin, EBITDA, debt-to-equity, current ratio, quick ratio, book value per share, free cash flow, cash flow margin, ROE, ROA, P/E, P/B, dividend yield)
• Calculate technical indicators (SMA, EMA, RSI, MACD, Bollinger Bands, volatility, ATR, Stochastic)
• Calculate period-over-period and annualized growth rates
• Backtest a moving-average crossover (Golden Cross / Death Cross) strategy
• Visualize price data with line and candlestick charts
• Plot financial metrics over time
• Compare multiple tickers side-by-side with optional normalization
• Plot trading signals and opportunities
• LLM-friendly, JSON-serializable outputs

Available Tools

Data Retrieval

Tool	Description
retrieve_current_price	Current price and related info for a ticker
retrieve_stock_data	Historical OHLCV data for one or more tickers
retrieve_financial_statements	Annual income statement, balance sheet, or cash flow
retrieve_ticker_info	General info and fundamentals for a ticker
retrieve_dividends	Dividend history for a ticker
retrieve_splits	Stock split history for a ticker

Metrics & Analysis

Tool	Description
mcp_calculate_financial_metrics	Compute a financial ratio or metric from statement data
mcp_calculate_technical_indicators	Compute SMA, EMA, RSI, MACD, Bollinger Bands, volatility, ATR, or Stochastic
mcp_calculate_growth_rates	Period-over-period and annualized growth rates for a time series
mcp_backtest_strategy	Backtest a MA crossover strategy; returns total return, Sharpe ratio, max drawdown, win rate, and individual trades

Visualization

Tool	Description
mcp_plot_price_chart	Line or candlestick price chart from OHLCV data
mcp_plot_financial_metric	Plot a financial metric over time
mcp_plot_comparison_chart	Compare multiple tickers or metrics; supports normalization to a common base of 100
mcp_plot_trading_opportunities	Price chart overlaid with MA crossover buy/sell signals

Note: All visualization tools automatically save charts to the data/ directory and open them with your system's default image viewer (macOS, Linux, and Windows are supported).

Installation

1. Clone the repository
   复制

   git clone https://github.com/Nishaant-Soni/Tearsheet.git
   cd Tearsheet
   

2. Install dependencies (requires uv)
   复制

   uv sync
   

How to Run

1. Configure your LLM client (Claude, Cursor, etc.) to connect to the MCP server and call the available tools.

   Go to Claude settings → Developer → Edit Config and add a new MCP server entry:

   复制

   {
     "mcpServers": {
       "tearsheet": {
         "command": "uv",
         "args": [
           "--directory",
           "/path/to/Tearsheet",
           "run",
           "main.py"
         ]
       }
     }
   }
   

   tearsheet should now be listed in your Claude tools.

2. Start talking with your trading assistant!

   Example queries:

   • Compare the stock prices of Nvidia and AMD in the past month.
   • How's Tesla stock looking in the past 3 months?
   • Plot the trading opportunities of Microsoft in the past 3 months.
   • What is the ROE and P/E ratio of Apple?
   • Show me the dividend history of Coca-Cola.
   • Backtest a moving-average crossover strategy on TSLA for the past year.
   • Plot a candlestick chart for NVDA over the last 30 days.
   • Compare AAPL, MSFT, and GOOGL returns over the past 6 months on a normalized chart.