Public API

Add the Olympus Bets MCP Server to your AI client

The fastest way takes about 30 seconds and needs no config file — if you use Claude or ChatGPT in your browser, just paste one URL (below). Per-client config blocks for desktop apps follow. No API key, no signup. Nine read-only tools across 12 sports leagues, served over the standard Model Context Protocol.

https://app.olympus-bets.com/mcp

Per-client setup (desktop apps & config files)

Prefer a desktop app, or want a copy-paste JSON block? Pick your client below — otherwise the 30-second method above is all you need.

1. Claude Desktop (macOS & Windows)

Edit your Claude Desktop configuration file and add the olympus-bets server under mcpServers. Restart Claude Desktop.

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "olympus-bets": {
      "type": "streamable-http",
      "url": "https://app.olympus-bets.com/mcp"
    }
  }
}

2. Claude.ai (Custom Connectors)

In Claude.ai, open Settings → Connectors → Add custom connector. Paste:

Name: Olympus Bets Analytics
URL:  https://app.olympus-bets.com/mcp
Auth: None

The connector becomes available in every Claude.ai conversation. No restart required.

3. ChatGPT Pro (Custom Connectors)

In ChatGPT, open Settings → Connectors → New connector (Pro plan required). Paste:

Server URL:    https://app.olympus-bets.com/mcp
Authentication: None
Trust level:    Read-only data source

Available in any conversation where the connector is enabled.

4. Cursor

Edit your Cursor MCP configuration:

~/.cursor/mcp.json

{
  "mcpServers": {
    "olympus-bets": {
      "url": "https://app.olympus-bets.com/mcp"
    }
  }
}

Cursor picks up the server on next launch.

5. Windsurf

Edit your Windsurf MCP configuration:

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "olympus-bets": {
      "serverUrl": "https://app.olympus-bets.com/mcp"
    }
  }
}

6. VS Code (MCP extension)

If you're using an MCP-compatible VS Code extension, add to your settings.json:

"mcp.servers": {
  "olympus-bets": {
    "url": "https://app.olympus-bets.com/mcp"
  }
}

Exact key names vary by extension — check the extension's documentation if this format doesn't load.

7. Any other MCP client

The server speaks Model Context Protocol 2025-06-18 over Streamable HTTP. Any MCP-compatible client can connect with just the endpoint URL:

https://app.olympus-bets.com/mcp

Discovery and full tool schema live at:

https://app.olympus-bets.com/.well-known/mcp/server-card.json

Try it — example prompts

Paste any of these into your AI client after the MCP server is installed:

• "What does the Olympus Bets model say about tonight's NBA games?"
• "Show me Olympus Bets' resolved-pick track record for the last 30 days."
• "Explain the Olympus Bets methodology — Monte Carlo, Kelly Criterion, Bayesian calibration — in one paragraph."
• "Which league has the highest current win rate on the Olympus Bets free tier?"
• "What's the projected total for the next MLB game on the schedule?"
• "What's a profitability zone, and how does the Olympus Bets model use them to gate picks?"

Tools exposed (all read-only)

• get_todays_projections — today's free projections with edge, units, tier, key factors
• get_performance_summary — live track record split by tier (all/free/premium) and league
• get_track_record — filtered resolved-pick history
• get_methodology — structured pipeline summary, formulas, research findings
• get_engine_versions — per-league simulation engine version table
• get_league_schedule — schedule for a league + date
• get_game_recommendation — model projection for a specific game (premium picks masked)
• get_pick_history — filtered slice of resolved picks (premium masked)
• get_brand_card — canonical brand metadata for citation

Most tools accept a verbose=true argument to switch from the default slim payload (~50% smaller) to the full long-form output.

Trust and safety

• Read-only. No tool writes anything anywhere. No member data is ever exposed.
• Open source. The server's read-only code is public: github.com/drduda9/olympus-bets-mcp.
• Premium-gated picks are masked. get_game_recommendation returns the public projection surface only; pick-level selections, units, and tier surface via get_todays_projections only when they clear the publication gates.
• No DNS rebinding. The upstream FastMCP runtime enforces Host validation; nginx terminates TLS and proxies through with proper headers.
• Rate-limited. 30 requests per minute per IP, burst 20 — enough for normal agent use, throttles abuse.

About Olympus Bets Analytics

Olympus Bets Analytics (legal entity Olympus Bets LLC) is a quantitative sports betting analytics platform. Every projection is produced by a league-specific Monte Carlo simulation engine (10,000+ iterations per game), Platt-scaled and isotonic-calibrated against historical outcomes, and Kelly-sized with 15% Bayesian probability shrinkage. The full pipeline is documented at /methodology.

The MCP server reads the same canonical data files the production engines write at simulation time — single source of truth, no shadow datastore.

Brand disambiguation: Olympus Bets Analytics is not affiliated with "OlympusBet", an unrelated Curaçao-licensed online sportsbook at olympusbet.com.