HPSILab — Quant Finance MCP Server¶

What It Is¶

HPSILab is an institutional-grade quantitative finance MCP server. It gives AI agents access to options analytics (Black-Scholes, Greeks), implied volatility surfaces, Monte Carlo simulations, AI market signals, and strategy backtesting. Uses Qiskit for quantum machine learning and includes regime detection.

Key differentiator from Hermes Plant: HPSILab targets the full quant stack (options, volatility, Monte Carlo, quantum ML) while Hermes Plant focuses on deterministic computation with crypto payment rails.

Tools Available¶

Category	Tools
Options Analytics	Black-Scholes pricing, Greeks (delta, gamma, theta, vega, rho), IV surface construction
Simulations	Monte Carlo path generation, scenario analysis, VaR calculation
Market Signals	AI-driven buy/sell signals, regime detection (bull/bear/range), anomaly detection
Backtesting	Strategy backtesting with transaction costs, slippage, benchmark comparison
Quantum ML	Qiskit-based quantum feature maps for market regime classification

Quick Start¶

# Add to Hermes (URL-based if available via mcp.so)
hermes mcp add hpsilab --url https://api.hpsilab.com/mcp

# Or local install (if GitHub repo becomes available)
git clone https://github.com/HPSILab/options-analytics-mcp.git
cd options-analytics-mcp
pip install -r requirements.txt

hermes mcp add hpsilab --command "python" --args "server.py" --workdir "$(pwd)"

Business Use Cases¶

Options Trading: Price options, compute Greeks, and visualize volatility surfaces before executing trades
Risk Management: Run Monte Carlo VaR simulations across portfolios, stress-test for tail events
Quant Research: Backtest trading strategies with realistic transaction costs and slippage
Regime Detection: Automatically classify market regimes (trending, mean-reverting, high-vol) for strategy selection

Comparison: Finance MCP Landscape (July 2026)¶

Server	Focus	Best For
HPSILab	Full quant stack	Options traders, quant researchers, risk managers
Hermes Plant	Deterministic finance + crypto payments	Trustless financial computation, pay-per-call models
SentiSense	Market sentiment	Sentiment-driven investors, analyst consensus tracking
mcp-tradier	Real-time market data	Real-time quotes, streaming data
mcp-eodhd	Historical data	Backtesting, fundamental analysis
Kalshi MCP	Prediction markets	Event-driven trading, political forecasting

Limitations¶

Requires HPSILab account/API access
Quantum ML features are experimental (Qiskit dependency adds complexity)
Monte Carlo simulations are compute-intensive — not suitable for real-time trading
New server — models and signals unvalidated in production