Skip to main content
Glama
Prediction Market MCP

Prediction Market MCP

by kingmadellc

Prediction Market MCP

A legacy TypeScript MCP client that connects an MCP host to a separate prediction-market research API. Maintained by King Made as the source and support destination for the historical npm package @kingmadellc/prediction-market-mcp@1.0.1.

Maintenance mode · experimental client · no automatic payments.

How it works

flowchart LR
    A[Your MCP host] -->|stdio| B[This client]
    B -->|HTTPS request| C[Separate hosted API]
    C --> D{Response}
    D -->|Success| E[JSON research output]
    D -->|HTTP 402| F[Payment-required information]
    E --> A
    F --> A

The repository contains the client, not the hosted service. It does not contain a wallet, sign a payment, place trades, or retry an HTTP 402 response with payment authorization. Requests send the supplied tool arguments to the external API configured in src/index.ts.

Related MCP server: proofmarket-mcp

Available tools

Tool

Client request

Purpose

prediction_market_estimate

GET /api/estimate

Request a probability estimate

prediction_market_size

POST /api/size

Request a sizing calculation

prediction_market_scan

POST /api/scan

Submit markets for ranking

prediction_market_ensemble

POST /api/ensemble

Request combined analysis

prediction_market_info

GET /

Read service information

Service availability, paid execution, and previously advertised performance figures have not been validated end to end. The client includes historical price strings; use the actual service response for current requirements rather than treating those strings as an offer.

Use the historical npm release

Requires Node.js 18 or newer, as declared in the package metadata. Repository CI uses Node.js 22.

For an MCP host that accepts a command and arguments, configure:

{
  "mcpServers": {
    "prediction-market": {
      "command": "npx",
      "args": ["-y", "@kingmadellc/prediction-market-mcp@1.0.1"]
    }
  }
}

Alternatively, install the command globally:

npm install -g @kingmadellc/prediction-market-mcp@1.0.1

The published npm version is historical. Commits to this repository do not update it automatically. The stdio process is designed to be launched by an MCP host, not opened as a website.

Develop locally

git clone https://github.com/kingmadellc/prediction-market-mcp.git
cd prediction-market-mcp
npm ci --ignore-scripts
npm run build
npm pack --dry-run

Point your MCP host at node with the absolute path to the resulting dist/index.js when testing your checkout. npm run dev starts the TypeScript entry point for development.

Path

Purpose

src/index.ts

Tool definitions, API routing, and stdio server

dist/index.js

Compiled CLI entry point

package.json

Command, package metadata, and scripts

.github/workflows/build.yml

Build, package inspection, and dependency audit

CI does not call the service or publish to npm. No live financial transaction is part of repository verification.

Support and license

Report reproducible client bugs, including the tool name and sanitized error output. Do not include credentials, wallet keys, account exports, or private financial data. No response-time commitment is offered for this legacy client.

MIT license, matching the existing package declaration. Preserve the license notice when redistributing the client. The separate hosted API has its own service requirements.

Available Tools

5 tools
prediction_market_ensembleB

Full prediction market analysis pipeline — Kalshalyst probability estimation + Xpulse social signal detection + ensemble weighting (0.75/0.25 Kalshalyst/Xpulse) + Kelly position sizing + market filter. Includes the complete prediction workflow with experimental estimation and signal detection. Costs $0.15 via x402 micropayment (USDC on Base).

Inputs: question (required), market_price (required), context, category, +1 more.

prediction_market_estimateA

Experimental probability estimate for a prediction market question. Historical performance claims have not been independently reproduced for this client. Returns probability estimate, confidence, reasoning, and recommended trade direction. Costs $0.05 via x402 micropayment (USDC on Base).

Inputs: question (required), market_price (required), context, category.

prediction_market_infoA

Get information about the Prediction Market Analyst service — available endpoints, pricing, methodology overview, and how x402 payment works. Free, no payment required.

Inputs: no named parameters.

prediction_market_scanB

Scan and rank multiple prediction markets by tradeable edge. Applies the market filter (skip rules remove low-quality markets, boost rules amplify high-edge categories). Returns ranked opportunities with Kelly sizing and filter status. Costs $0.10 via x402 micropayment (USDC on Base).

Inputs: markets (required).

prediction_market_sizeA

Kelly-optimal position sizing for a prediction market trade. Uses configured Kelly parameters (α=0.75, conf_exp=1.0); no performance improvement is guaranteed. Applies market filter rules (skip fed/ultra-low/short-duration, boost policy/tech/markets). Costs $0.08 via x402 micropayment (USDC on Base).

Inputs: confidence (required), market_price (required), estimated_prob (required), bankroll, +1 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 5 tools

    Disambiguation4/5

    Each tool has a distinct primary purpose: estimate for single-question probability, size for position sizing, scan for ranking multiple markets, ensemble for a full pipeline, and info for service details. However, ensemble overlaps with estimate and size by including them, which could confuse agents choosing between atomic and composite options.

    Naming Consistency5/5

    All tools follow the same snake_case pattern with a consistent 'prediction_market_' prefix followed by a clear action or noun (estimate, size, scan, ensemble, info). This makes the set highly predictable and readable.

    Tool Count5/5

    Five tools is well-scoped for a prediction market analysis service, with each tool earning its place by covering a distinct step in the workflow from information to estimation, sizing, scanning, and full pipeline execution.

    Completeness4/5

    The surface covers the core analysis lifecycle: service info, single-market estimation, position sizing, multi-market scanning, and an ensemble pipeline. Minor gaps exist around raw market data retrieval or trade execution, but agents can work around these for the stated analysis purpose.

    Maintenance

    ActivityStale
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      B
      maintenance
      Quant tools for AI agents on Kalshi & Polymarket. 34 tools, 27 free with no key: EV, Kelly sizing, Bayes updates, odds conversion, base-rate gaps, combo grading, Fed odds, NFL and NHL models, 2026 Senate race odds, Kalshi weather board and Kalshi 15-minute markets + perps (live board + liquidation estimates). Pro adds commodity + NFL edge models, mispricing scanner and alerts.
      44
      56 npm
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      Exposes a machine-callable proof-market surface for agents to submit proof requests, monitor market state, compare coordination regimes, and compose multi-step workflows with no human in the loop.
      5
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Provides prediction-market data as MCP tools, including live prices, keyword search, probability history, volume spikes, arbitrage gaps, and resolutions. Pay per call in USDC on Base with no API key or signup.
      29 npm
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Prediction-market MCP server for Kalshi and Polymarket: market search, live order books and quotes, each venue's fee schedule, fee-aware no-arbitrage checks across related event contracts, and paper orders that walk the book.
      14
      172 PyPI
      569
      Apache 2.0