matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun)

Registre officiel Jeux, sport et divertissement

One row per game, matched across Kalshi, Polymarket US/Intl and Predict.fun. Mapping only.

Installation

Ajoutez ceci à la configuration MCP de votre client (Claude Desktop : claude_desktop_config.json ; Cursor : .cursor/mcp.json ; VS Code et les autres utilisent la même forme « mcpServers »).

Remote (streamable-http)
Claude Code
claude mcp add --transport http matchwire-win-mapped-prediction-markets- 'https://api.matchwire.win/mcp' --header 'x-api-key: <matchwire.win API key (mw_...). Required: every tool result is narrowed to the key'\''s plan exactly like the HTTP API. Keys are issued at https://matchwire.win/#pricing — plans start at $99/month, 14-day trial on Starter.>'
Configuration JSON (Claude Desktop, Cursor, VS Code)
{
    "mcpServers": {
        "matchwire-win-mapped-prediction-markets-": {
            "type": "http",
            "url": "https://api.matchwire.win/mcp",
            "headers": {
                "x-api-key": "<matchwire.win API key (mw_...). Required: every tool result is narrowed to the key's plan exactly like the HTTP API. Keys are issued at https://matchwire.win/#pricing — plans start at $99/month, 14-day trial on Starter.>"
            }
        }
    }
}

Documentation

Tirée du README du projet sur GitHub, actualisée chaque semaine. Les liens ouvrent GitHub.

Sommaire
  1. matchwire.win
  2. Install
  3. Quickstart
  4. API
  5. Errors
  6. Documentation
  7. Python
  8. License

matchwire.win

Mapped prediction-market data for Node.js and Python. One client, no dependencies: matchwire.win returns one row per game, matched across Kalshi, Polymarket US, Polymarket International and Predict.fun, with each venue's own market id left intact — the pieces you need to compare a Kalshi API market with its Polymarket API twin without owning a reconciliation job.

Market types are matched too: winners, spreads and handicaps, totals and props. The line is part of the match, so over 45.5 is never confused with over 48.5.

Mapping only — no prices, no order books. Prices come from each venue; this library tells you which market on which venue is the same game.

import { Matchwire } from 'matchwire.win';

const mw = new Matchwire({ apiKey: process.env.MATCHWIRE_API_KEY });

const { rows, seq } = await mw.rows({ sport: 'nfl', venue: ['kalshi', 'poly'] });
const game = rows.find((r) => r.listings.length > 1);
console.log(game.name, game.listings.map((l) => `${l.venue}:${l.marketId}`));
  • Zero dependencies. One src/ directory on the global fetch. Nothing installs alongside it.
  • Node.js 18+, ESM and CommonJS, hand-written TypeScript declarations.
  • Typed errors — a 429 is distinguishable from a 401 invalid_key in code, not by string matching.
  • Deltas and streaming — since=<seq> for changed rows and tombstones, or the same updates over SSE.
  • MCP included — the same tools an AI agent gets from https://api.matchwire.win/mcp.
  • MIT licensed.

Install

npm install matchwire.win      # Node.js — zero dependencies
pip install matchwire.win      # Python 3.9+ — standard library only

Get a key at matchwire.win — plans start at $99/month with a 14-day trial on Starter.

Quickstart

Save as quickstart.mjs and run node quickstart.mjs (Needs MATCHWIRE_API_KEY in the environment):

import { Matchwire } from 'matchwire.win';

const mw = new Matchwire({ apiKey: process.env.MATCHWIRE_API_KEY });

const status = await mw.status();
console.log(`${status.store.events} games, ${status.store.bound} bound listings`);

const { sports } = await mw.sports();
console.log(sports.slice(0, 5).map((s) => `${s.sport} ${s.events}`));

const { rows } = await mw.rows({ sport: 'nfl', limit: 3 });
for (const row of rows) {
  console.log(row.name, row.status, row.listings.map((l) => l.venue).join('+'));
}

The repository ships the same thing with formatting and error handling: node examples/quickstart.mjs.

API

MethodEndpointReturns
status()GET /api/v1/statusService and per-venue health, current seq, coverage counts
sports()GET /api/v1/sportsSports with game counts, plus group ids (esports, american-football, …)
rows(params)GET /api/v1/rowsOne row per game with its per-venue listings
row(id)MCP get_rowOne game by its mapped id
stream(params)GET /api/v1/pushThe same updates as Server-Sent Events
callTool(name, args)POST /mcpAny tool the MCP endpoint advertises
tools()POST /mcpThe tool list the endpoint advertises
Filters

sport, venue, type, league, status accept a value or a comma list; they combine. since=<seq> switches to a delta (changed rows plus tombstones).

  • venue= kalshi · poly (Polymarket US) · polyintl (Polymarket International) · predictfun
  • type= winner · total · handicap · map_winner · score · event
  • status= scheduled · live · ended
  • sport= a sport name (baseball, not mlb) or a group id from sports()
// Everything that changed since the last call, plus tombstones for pruned games.
let seq = 0;
const first = await mw.rows({ sport: 'esports', limit: 5000 });
seq = first.seq;

const delta = await mw.rows({ since: seq });
console.log(delta.rows.length, 'changed', delta.tombstones?.length ?? 0, 'gone');

// Stream instead of polling.
for await (const frame of mw.stream({ since: seq })) {
  console.log(frame.event, frame.data.rows?.length);
  if (Date.now() - started > 10_000) break;   // or AbortController
}
Mapping, not prices

Every listing carries the venue's own marketId, a tier (exact, normalized, fuzzy) and a confidence, plus the exact request that reads that market from the venue. Use the mapped row to decide what to compare, then read the prices from each venue with the source you already have.

Agents and MCP

POST https://api.matchwire.win/mcp speaks JSON-RPC 2.0 with your key in the same x-api-key header. Tools: list_sports, find_rows, get_row, get_venue_market, report_issue, service_status. Read-only; every tool result is narrowed by your key's plan exactly like the HTTP API.

import { Matchwire } from 'matchwire.win';
const mw = new Matchwire({ apiKey: process.env.MATCHWIRE_API_KEY });

console.log((await mw.tools()).map((t) => t.name));
const row = await mw.row('ev:nfl:arizona-cardinals:denver-broncos:2026-10-25');
console.log(row.listings.map((l) => l.venue));

Plain text for coding assistants: matchwire.win/llms.txt · MCP reference: matchwire.win/docs/mcp/

Errors

import { MatchwireError } from 'matchwire.win';

try {
  await mw.rows({ sport: 'nope' });
} catch (error) {
  if (error.isRateLimited) await sleep(error.retryAfterMs);   // 429
  else if (error.isAuthError) /* key missing or revoked */ ;
  else if (error.isPlanError) /* outside the plan's sports/venues */ ;
  else throw error;
}

Documentation

Python

from matchwire import Matchwire

mw = Matchwire(api_key=os.environ["MATCHWIRE_API_KEY"])
rows = mw.rows(sport="nfl", venue=["kalshi", "poly"])["rows"]
print(rows[0]["name"], [l["venue"] for l in rows[0]["listings"]])

See python/README.md. Standard library only, Python 3.9+.

License

MIT. Not affiliated with Kalshi, Polymarket or Predict.fun; venue names are used to say what is mapped.

Questions fréquentes

Qu'est-ce que matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun) ?

One row per game, matched across Kalshi, Polymarket US/Intl and Predict.fun. Mapping only.

Comment installer matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun) ?

Il fonctionne comme serveur distant. Dans Claude Code : claude mcp add --transport http matchwire-win-mapped-prediction-markets- 'https://api.matchwire.win/mcp' --header 'x-api-key: <matchwire.win API key (mw_...). Required: every tool result is narrowed to the key'\''s plan exactly like the HTTP API. Keys are issued at https://matchwire.win/#pricing — plans start at $99/month, 14-day trial on Starter.>'. Pour les autres clients, collez la configuration JSON de cette page dans leurs réglages MCP.

Faut-il une clé API pour matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun) ?

Oui. Il attend x-api-key, à renseigner dans sa configuration.

matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun) est-il open source ?

Oui. Son code est sur GitHub sous licence MIT.

Qui maintient matchwire.win — Mapped Prediction Markets (Kalshi, Polymarket, Predict.fun) ?

Il est publié par MatchWire-Win. AI.tn le référence sans le maintenir ni le recommander.

Issu du registre MCP officiel. Être listé n'est pas une recommandation : vérifiez le code et les permissions d'un serveur avant de lui donner accès à vos données.