Skip to main content

Connect your AI to Rising Gamma

You already ask Claude or ChatGPT about the markets, and today they answer from whatever they happened to read on the web. With Pro + Connections they answer by running Rising Gamma: the same Monte Carlo, the same portfolio optimizer, the same option pricing you would open on this site, called from inside the chat, with the numbers and with what those numbers mean. You keep the assistant you already use. It gains the platform.

The open standard that makes this possible is called MCP (Model Context Protocol). Setting it up is one command, or one screen in your client's settings, once. Everything below is that setup, client by client.

With the Pro + Connections plan, Rising Gamma becomes a remote MCP server your AI clients call directly: Claude on the web, Claude Code and LLM agents in your terminal (CLI), Cursor, ChatGPT and any MCP-capable client. Every analytics tool the platform offers (Monte Carlo, Markowitz, delta hedging, backtesting, Black-Scholes, correlations, COT, fundamentals and more) is exposed as a tool, and every output ships with its context: what each input means, how results are computed, and how to read every data series (the same series the platform charts).

What a session looks like

The point is not that your AI can fetch a number. It is that the number arrives with the instructions to read it, in the same reply.

You

I'm retiring on 1,000,000 and I want to take 40,000 a year for 30 years. Does that hold up?

Tool calls
search_assets("S&P 500 ETF")  ->  SPY

montecarlo_planning(
  ticker: "SPY",
  initial_capital: 1000000,
  annual_withdrawal: 40000,
  time_horizon_years: 30,
  inflation_rate: 0.025
)
Your AI
success_probability
0.87
median_final_wealth
3,180,000
percentile_10th
210,000
percentile_90th
9,850,000

87% of the 500 simulated paths still had capital at year 30. The withdrawal is inflated 2.5% a year, so the 40,000 you take in year 1 is about 82,000 by year 30, and that is what sinks the weak paths: the 10th percentile finishes at 210,000, one bad decade away from zero. Returns are drawn from a Johnson SU fit rather than a normal curve, so the crash years that decide those paths are not smoothed away. Want me to rerun it at 35,000 a year, or on your own portfolio instead of SPY?

Example session. The tool names, the inputs and the fields in the answer are the real ones; the figures depend on the asset and the day you run it.

The endpoint

https://api.risingamma.com/mcp

Authorization uses OAuth with your Rising Gamma account: the first time a client connects, a sign-in page opens, approve it once and revoke it anytime from your account area (AI access section). MCP access requires the Pro + Connections plan.

Claude Code / terminal (CLI)

From any terminal with Claude Code installed:

claude mcp add --transport http rising-gamma https://api.risingamma.com/mcp

The browser opens for the one-time sign-in; after that, ask Claude things like “run a Monte Carlo financial plan on my portfolio”, it will call the platform's tools directly.

claude.ai (web)

  1. Settings → Connectors → Add custom connector
  2. Paste the endpoint URL above and confirm
  3. Sign in with your Rising Gamma credentials when prompted

Cursor and generic MCP clients

Add the server to your mcp.json (or equivalent config):

{
  "mcpServers": {
    "rising-gamma": { "url": "https://api.risingamma.com/mcp" }
  }
}

ChatGPT

Enable Developer mode (Settings → Connectors), then add the endpoint as a connector. Note: plain (non-developer) ChatGPT connectors expect search/fetch-style tools, Rising Gamma's search_knowledge and get_knowledge double as those, but full analytics tools require developer mode. Capabilities vary by ChatGPT plan.

What your AI can call

  • 30+ analytics tools, such as options pricing & strategies, Monte Carlo (strategy robustness, financial planning), Markowitz optimization, delta hedging, distribution/volatility/correlation/cycle analysis, backtesting, COT, SEC financials, 13F, insider trading, IPOs and more.
  • Workflow tools: search_assets (resolve tickers first), validate_portfolio (weights must sum to 100%), get_asset_overview (prices, volatility & standing as context before any analysis).
  • Knowledge tools: list_knowledge, get_knowledge, search_knowledge, get_platform_info: the platform's knowledge base (40+ topics), pricing FAQ, terms and feature-page catalog, so the AI always has the context behind each number.
  • get_quota: remaining calls at any moment.

Usage quota

The Connected plan includes a dedicated quota of up to 1,000 tool calls a day per account, sized for real agentic workflows. Requests that just prepare an analysis, resolving tickers, validating a portfolio, reading an asset's context or the built-in knowledge base, don't count against it; only the analyses themselves do. A short-term per-minute limit keeps the service responsive. Your AI reads the figures in force live with get_quota, and when a limit is reached the server replies with a structured message telling the client exactly when to resume, so your AI handles it gracefully.

Good to know

  • Outputs are statistical analyses, not investment advice; your AI may transform or interpret them (see the Terms of Service).
  • Tickers in the dataset are exchange-qualified (e.g. APC.F): always resolve them with search_assets first.
  • You can revoke any connected client from Space → AI access; its tokens stop working within the hour.
Get Pro + Connections