QuantData MCP Server

THIS IS AN UNOFFICIAL MCP SERVER I AM NOT AFFILIATED WITH QUANTDATA IN ANY WAY

MCP server that gives AI agents (Claude Code, Claude Desktop, etc.) access to real-time and historical options market data from QuantData.

📖 New to this? Start with the Getting Started Guide for a step-by-step walkthrough.

Supports any optionable ticker — SPX, SPY, QQQ, AAPL, TSLA, and more. Not just 0DTE.

Available data: GEX/DEX/CEX/VEX exposure walls, exposure term structure, net drift, max pain (instantaneous + per-expiration), IV rank, IV skew, IV term structure, volatility drift (ARV vs IV), trade side statistics, open interest (per strike, per expiration, over time, day-over-day change), net flow, consolidated + unconsolidated order flow, contract OHLCV, and contract statistics.

Documentation: see GETTING_STARTED.md for a click-by-click walkthrough.

Quick Start

1. Install

You need Python 3.11+ installed. Check with python3 --version.

• Mac: brew install python (or download from python.org)

• Windows: Download from python.org (check "Add to PATH" during install)

Install from PyPI:

# With pip
pip install quantdata-mcp

# With uv (faster)
uv pip install quantdata-mcp

# Or install as a global tool with uv
uv tool install quantdata-mcp

Don't have uv? Install it with curl -LsSf https://astral.sh/uv/install.sh | sh (Mac/Linux) or irm https://astral.sh/uv/install.ps1 | iex (Windows). It's a faster alternative to pip.

Want the bleeding edge? Install from GitHub: pip install git+https://github.com/zzulanas/quantdata-mcp.git

2. Get Your Credentials

You need two values from your QuantData account. Open your browser:

1. Go to v3.quantdata.us and log in

2. Open DevTools (F12 or right-click → Inspect) → Network tab

3. Refresh the page

4. Click on any chart or page on QuantData — you'll see API requests appear

5. Click any request to core-lb-prod.quantdata.us, or filter by /api in the top

6. In the Request Headers, find and copy:

   • authorization — your auth token (starts with eyJ...)

   • x-instance-id — your instance ID (a UUID like xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx)

Should look like these:

3. Run Setup

quantdata-mcp setup \
  --auth-token "eyJhbGci..." \
  --instance-id "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"

This creates a dedicated page on your QuantData account with 19 data tools and saves your config to ~/.quantdata-mcp/config.json.

Upgrading from a previous install? Just upgrade the package — the server auto-registers any new tool definitions (e.g. the 8 Tier-1 additions in this release) on first start. Existing tool IDs and page layout are preserved. Re-running quantdata-mcp setup is no longer required when a release adds new tools; only run it again if you need to refresh an expired auth token.

4. Add to Claude

Claude Code

Add to your project's .mcp.json (or global ~/.claude/mcp.json):

{
  "mcpServers": {
    "quantdata": {
      "command": "quantdata-mcp",
      "args": ["serve"]
    }
  }
}

Restart Claude Code. You should see quantdata in your MCP servers.

Claude Desktop

Add to your Claude Desktop config file:

• Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "quantdata": {
      "command": "quantdata-mcp",
      "args": ["serve"]
    }
  }
}

Note: quantdata-mcp must be on your system PATH. If it's not found, use the full path:

which quantdata-mcp   # find the path

{ "command": "/Users/you/.local/bin/quantdata-mcp", "args": ["serve"] }

Or use uvx to run without worrying about PATH:

{
  "command": "uvx",
  "args": ["--from", "git+https://github.com/zzulanas/quantdata-mcp.git", "quantdata-mcp", "serve"]
}

Restart Claude Desktop. The QuantData tools will appear in your tool list.

Available Tools

Market Overview

Exposure (Greeks)

Premium Flow

Order Flow & Trade Stats

qd_get_order_flow is the most filter-rich tool — it now exposes the full QuantData order-flow filter set (40+ flat kwargs grouped into bool flags, GTE/LTE thresholds, greek thresholds, and multi-select lists). Example — bullish opening-position sweeps in tech with $10K+ premium:

qd_get_order_flow(
    ticker="SPY", is_unusual=True, is_opening_position=True,
    sentiment_type=[SentimentType.BULLISH], min_premium=10000,
    trade_type=["AUTO"], sector=["TECHNOLOGY"], last_n=20,
)

Volatility & Pricing

Open Interest & Max Pain

Common Parameters

All tools accept these parameters for ticker/date control:

Filter Parameters

Tools that support filtering accept these optional parameters:

Example Usage

Ask Claude things like:

• "What are the biggest GEX walls right now?"

• "Show me yesterday's DEX walls at 10:30 AM"

• "Pull up the trade side stats — are puts or calls more aggressive today?"

• "Compare the GEX profile at open vs close for last Thursday"

• "Show me AAPL gamma exposure for the April 17 monthly expiration"

• "What's the OTM-only net drift for SPX today?"

• "Show me the order flow — just calls with premium over $50K"

• "Get the OHLCV for the SPX 6600 call today"

• "What's the IV rank with a 30-day lookback?"

• "Show the GEX term structure across all expirations"

Multi-Ticker Support

All tools work with any optionable ticker. Just pass ticker="AAPL" (or whatever symbol).

Important: SPX, SPY, and QQQ have daily expirations (0DTE works by default). For equity options like AAPL or TSLA, you must set expiration_date to a valid expiration (e.g. monthly 3rd Friday) or you'll get empty data.

> Show me TSLA GEX walls for the April 17 monthly
> qd_get_exposure_by_strike(ticker="TSLA", expiration_date="2026-04-17")

Historical Data + Time Scrubber

All tools support historical analysis. Set context once via qd_set_page_date and subsequent calls inherit it (the page filter is sticky as of v0.3.0):

> Switch to RKLB on 2026-05-08, then show me the gamma walls

> qd_set_page_date(date="2026-05-08", ticker="RKLB", expiration_date="2026-05-15")
> qd_get_exposure_by_strike()    # inherits RKLB / 5/8 / 5/15
> qd_get_max_pain()                # still RKLB / 5/8 / 5/15

Persistent intraday scrubber — set a tool to a specific moment of the trading day and the chart in your QuantData browser updates live:

> Show me the GEX walls at 11:00 AM yesterday

> qd_set_tool_time("exposure_by_strike", "11:00")    # accepts "9:30", "1:30 PM", "16:00"
> qd_get_exposure_by_strike()                         # data scrubbed to 11:00 AM
> qd_reset_to_live("exposure_by_strike")              # back to most recent

Note: date must be a valid trading day (not weekends or market holidays).

Filter Groups (v0.3.0)

QuantData's filter groups are server-side, persistent, named filter sets that attach to tools. Once attached, the filter is AND'd onto every fetch — visible and editable in the QuantData web UI alongside what the LLM is doing.

> Save a filter group that excludes complex spreads, tied trades, and floor trades.
  Apply it to net_drift.

> qd_save_filter_group(
    name="clean_signal",
    conditions=[
      {"field": "IS_COMPLEX",   "op": "==", "value": False},
      {"field": "IS_TIED",      "op": "==", "value": False},
      {"field": "IS_FLOOR",     "op": "==", "value": False},
      {"field": "IS_CANCELLED", "op": "==", "value": False},
    ],
  )
> qd_apply_filter_group("net_drift", "clean_signal")

After this, every qd_get_net_drift(...) call automatically AND's those four conditions. Edit incrementally with qd_add_filter_clause, qd_update_filter_clause, qd_remove_filter_clause. Discover community filters with qd_search_public_filter_groups. Field/operator catalog: qd_list_filter_fields.

How It Works

QuantData doesn't have an official API. This server uses reverse-engineered REST endpoints from their web app. Each user has "tools" (chart widgets) on "pages" — the setup command creates a dedicated page with all 19 data types so the MCP server can query them.

Architecture:

Claude --> MCP (stdio) --> quantdata-mcp server --> QuantData REST API

Your credentials and tool IDs are stored locally at ~/.quantdata-mcp/config.json.

Commands

quantdata-mcp setup --auth-token <TOKEN> --instance-id <ID>  # One-time setup
quantdata-mcp serve                                           # Start MCP server (used by Claude)

Requirements

• Python 3.11+

• Active QuantData subscription

Troubleshooting

"Config not found" error: Run quantdata-mcp setup first.

Auth errors (401): Your token expired. Get a new one from the Network tab and re-run setup. Your existing page and tools will be reused:

quantdata-mcp setup --auth-token "NEW_TOKEN" --instance-id "SAME_ID"

Empty data: Make sure you have an active QuantData subscription and the market was open on the date you're querying. For non-index tickers (AAPL, TSLA), make sure you set expiration_date to a valid options expiration.

"No such file or directory" in Claude Desktop: Use the full path to quantdata-mcp (see step 4 above).

Contributing

Contributions are welcome — bug reports, new tools wrapping additional QuantData widgets, formatter improvements, and docs all help. See CONTRIBUTING.md for the dev environment setup, the recipe for adding a new MCP tool, and the conventions to follow when sending a PR.

License

MIT — see LICENSE.