HPSILab MCP

⭐ If you find HPSILab useful, please star the repository.

8 institutional-grade quantitative finance tools for AI agents and MCP clients.

Analyze stocks using AI forecasting, options positioning, implied volatility intelligence, Monte Carlo simulation, and strategy backtesting — each exposed as a dedicated MCP tool.

Official Remote MCP Endpoint

https://hpsilab.com/mcp

---

Quick Start

Which option should I use?

Option	Setup Time	Best For
Remote MCP (https://hpsilab.com/mcp)	Instant	Most users
Self-Hosted MCP Server	2–3 minutes	Developers
Enterprise Deployment	Custom	Organizations

Option 1 — Official Remote MCP Service (Recommended)

Connect directly to the official HPSILab MCP endpoint — no installation required, always up to date.

https://hpsilab.com/mcp

Option 2 — Open Source Self-Hosted MCP Server

git clone https://github.com/haiyunsky/hpsilab-mcp-server.git
cd hpsilab-mcp-server
pip install -r requirements.txt
python src/hpsilab_mcp_server/server.py

---

MCP Client Configuration

Get an API Key

Create a free account at hpsilab.com and generate an API key (hpsi_...) from the dashboard.

Cursor (Remote MCP)

{
  "mcpServers": {
    "hpsilab": {
      "url": "https://hpsilab.com/mcp",
      "headers": {
        "Authorization": "Bearer hpsi_your_key"
      }
    }
  }
}

Claude Desktop / Claude Code (via mcp-remote)

{
  "mcpServers": {
    "hpsilab": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://hpsilab.com/mcp",
        "--header",
        "Authorization: Bearer hpsi_your_key"
      ]
    }
  }
}

Self-Hosted (Cursor)

{
  "mcpServers": {
    "hpsilab": {
      "command": "hpsilab-mcp-server"
    }
  }
}

---

Available Tools

All tools accept a single symbol parameter: an exchange ticker in uppercase (e.g. "NVDA", "AAPL", "SPY").

analyze_stock

Full institutional-grade analysis — aggregates AI prediction, IV radar, options pressure, Monte Carlo, and backtesting into a single bull/bear verdict.

Use when: you need a holistic market view with confidence score and supporting evidence.

Returns: signal, confidence_score, bullish_factors, bearish_factors, summary

---

get_iv_radar

Implied volatility metrics: ATM IV, IV rank (0–100), IV percentile, risk reversal direction, and volatility regime.

Use when: you want to assess whether options are cheap or expensive, or identify the current vol regime.

Returns: atm_iv, iv_rank, iv_percentile, risk_reversal, volatility_regime

---

get_option_pressure

Options-market positioning and dealer-hedging pressure zones: max pain, gamma wall, expected move, and squeeze targets.

Use when: you need strike-level gravitational targets near expiration or want to size an expected-move trade.

Returns: max_pain, gamma_wall, expected_move, squeeze_target, expiry_date, pressure_zones

---

get_monte_carlo

10,000-path GBM Monte Carlo simulation over a 30-day horizon, calibrated with realized volatility and current IV.

Use when: you need a probabilistic price range, downside probability estimates, or volatility-adjusted scenarios.

Returns: mean_price, range_90, range_68, prob_above_spot, prob_10pct_drop, distribution

---

get_ai_prediction

Ensemble AI directional prediction (gradient-boosted trees + LSTM + quantum VQC) for the next session's move.

Use when: you want a data-driven up/down probability with per-model votes and market regime classification.

Returns: prediction, up_probability, confidence, model_votes, regime, signal_strength

---

get_equity_curves

Backtested equity curves and risk-adjusted metrics (Sharpe, Sortino, max drawdown, win rate) for standard quant strategies applied to the ticker.

Use when: you want historical performance context or need to compare strategy quality across tickers.

Returns: strategies[] — each with total_return, sharpe_ratio, max_drawdown, win_rate, equity_curve

---

generate_stock_research_report

Generates a structured markdown research note synthesizing all signal sources, suitable for sharing with investors.

Use when: a user asks for a "report" or "write-up" and needs a formatted narrative rather than raw JSON.

Returns: report (markdown string), generated_at

---

generate_stock_images

Returns public URLs for three charts: candlestick price chart, 3-D IV surface, and options flow heatmap. URLs expire after 24 hours.

Use when: a user asks to "see" or "visualize" a chart, or you want to embed visuals in a report.

Returns: price_chart_url, iv_surface_url, options_flow_url, expires_at

---

Example

# Quick directional verdict
analyze_stock("NVDA")

# Only need vol data
get_iv_radar("NVDA")

# Probabilistic price range
get_monte_carlo("NVDA")

Example analyze_stock response:

{
  "symbol": "NVDA",
  "signal": "Bearish",
  "confidence_score": 42,
  "bullish_factors": [
    "Monte Carlo range midpoint is above current spot.",
    "Option pressure leaves a meaningful upside weekly-high zone."
  ],
  "bearish_factors": [
    "AI prediction gives only a 34.2% probability of an up close.",
    "Max Pain sits below spot, suggesting downward expiry pin pressure.",
    "Risk reversal is put-heavy.",
    "All three AI models point down."
  ],
  "summary": "NVDA screens bearish with a 42/100 direction score."
}

---

Architecture

AI Client (Claude / Cursor / Windsurf / ...)
    ↓  MCP protocol
hpsilab-mcp-server  (this repo)
    ↓  HTTPS REST
HPSILab Quant API  (hpsilab.com)
    ↓
Quant Platform  (IV engine · ML models · Monte Carlo · Backtester)

---

Supported MCP Clients

Cursor · Claude Desktop · Claude Code · ChatGPT Agents · Cline · Roo Code · Windsurf · Continue · Any MCP-compatible client

---

Disclaimer

This software is provided for research and educational purposes only. Nothing contained in this project constitutes investment advice, financial advice, or a recommendation to buy or sell any security. Always perform your own due diligence before making investment decisions.

---

License

MIT License — Copyright (c) 2026 Haiyun Hu