Deskpricer

What is Deskpricer?

DeskPricer is a local HTTP pricing microservice for vanilla European and American equity options. It integrates with Excel via WEBSERVICE and FILTERXML (no VBA required) and is also available as an MCP server for AI agents like Cursor and Claude Desktop. It is designed for individual desk pricing and option analytics, not for public/server-style deployment.

How to use Deskpricer?

Install via pip: pip install deskpricer. For MCP use, add deskpricer-mcp to your agent’s MCP configuration (e.g., ~/.cursor/mcp.json or claude_desktop_config.json). For the HTTP API, run the standalone executable or python -m deskpricer.main; then call endpoints like GET /v1/greeks with query parameters. JSON is available via Accept: application/json or ?format=json.

Key features of Deskpricer

• Price + Greeks for single options and portfolios
• Implied volatility solver (Brent method via QuantLib)
• PnL attribution with delta, gamma, vega, theta, rho, vanna, volga
• XML-by-default output for seamless Excel integration
• Localhost-only binding (127.0.0.1) – no network exposure
• Four MCP tools: price_option, implied_volatility, pnl_attribution, portfolio_greeks

Use cases of Deskpricer

• Price vanilla options and Greeks in Excel without VBA or Bloomberg
• Back out implied volatility from market mid-prices
• Decompose option PnL into risk factor contributions
• Aggregate Greeks for multi-leg portfolios via the POST endpoint
• Use with AI agents (Cursor, Claude) for natural language option analytics

FAQ from Deskpricer

What are the runtime requirements?

The standalone executable (DeskPricer_v3.exe) requires no Python installation. Running from source needs Python, a C++ compiler for QuantLib (or a pre‑built wheel), and the deskpricer package.

Where does data live and how is it secured?

Data stays on your machine – the service binds to 127.0.0.1 only. No authentication, TLS, or rate limiting is implemented, as it is designed for local use only.

How do I get JSON output instead of XML?

Set the Accept: application/json header or append ?format=json to the URL. XML is used by default for compatibility with Excel’s FILTERXML.

What are the known limitations?

No finite‑difference engine; only analytic (European) and binomial (American) engines. Zero‑DTE options are floored to 1 calendar day. Engine/style mismatches (e.g., European with binomial) return an error.

Where are logs written?

Logs go to DESKPRICER_LOG_DIR (default: C:\ProgramData\DeskPricer\logs on Windows, ~/.local/share/deskpricer/logs elsewhere). Uses Python logging with JSON formatting and rotating files (10 MB, 5 backups).