Prediction Markets Scraper — MCP Server

One actor. Every market. AI-ready.

Scrape live prediction market events, prices, and liquidity across 13 exchanges — Polymarket, Kalshi, Limitless, Metaculus, Smarkets, and more — powered by the pmxtjs unified API.

Runs as an Apify actor for scheduled scraping and one-off runs. Because Apify connects native MCP integration, the data this actor collects is immediately consumable by AI agents like Claude and Cursor — no extra server required.

---

Supported Exchanges

Exchange	Region	Type
Polymarket	Global	Crypto
Polymarket US	US	Regulated
Kalshi	US	Regulated (CFTC)
Kalshi Demo	US	Demo
Limitless	Global	Crypto
Metaculus	Global	Forecasting
Smarkets	Europe	Exchange
Myriad	Global	Crypto
Probable	Global	Crypto
Opinion	Global	Social
Hyperliquid	Global	DeFi
Gemini Titan	US	Regulated
SuiBets	Global	Crypto

---

Apify Platform Capabilities

Running this as an Apify Actor gives you the full Apify infrastructure stack on top of the scraping logic:

• Scheduled runs — trigger on any cron schedule (hourly snapshots, daily digests, weekly archives) via the Apify Console or API.
• Webhooks — fire a POST to any endpoint when a run finishes, enabling downstream pipelines (Slack alerts, database writes, further actors).
• Dataset storage — all output lands in a structured Apify Dataset, queryable via REST API, exportable as JSON/CSV/XLSX, and viewable in the Console.
• Key-value store — the actor writes a run summary (record count, exchanges queried, finish time) to the default KV store after each run.
• Apify API — call the actor programmatically from any language, pass input overrides per run, and read results back without ever touching the Console.
• Actor-to-actor chaining — pipe this actor's dataset output directly into downstream actors (e.g. an LLM summariser, a notifier, a database sync actor) using Apify's dataset getItems API or the Actor.metamorph() pattern.
• MCP integration — Apify's native Model Context Protocol support lets Claude, Cursor, and other AI agents call this actor as a tool directly from the chat — no extra server or wrapper needed.
• Monitoring & alerts — set run-failure notifications and memory/timeout limits in the Console.
• Proxy support — plug in Apify's residential or datacenter proxies if any exchange starts rate-limiting headless requests.

---

Pricing

This actor uses pay-per-event billing — you only pay for what you actually scrape.

Event	Price	When charged
actor-start	$0.01	Once per run, before any fetching
prediction-market-event	$0.002	Per event scraped and saved to the dataset

Example costs:

Run	Records	Cost
50 events from 1 exchange	50	~$0.11
150 events from 3 exchanges	150	~$0.31
500 events, deduped cross-venue	500	~$1.01

The actor respects your maxTotalChargeUsd budget — if the limit is reached mid-run, it saves whatever was collected and exits cleanly without losing partial results.

Prices are set in the Publication tab of the Apify Console. The events emitted by the code are actor-start and prediction-market-event.

---

Input

Configure the actor in Apify Console or pass a JSON input file:

{
  "exchanges": ["Polymarket", "Kalshi", "Limitless"],
  "query": "US election 2026",
  "limit": 100,
  "status": "active",
  "category": "politics",
  "tags": ["US", "election"],
  "minVolume": 50000,
  "minLiquidity": 10000,
  "includeOutcomes": true,
  "includeSourceMetadata": false,
  "crossVenueDedup": true,
  "outputFormat": "events"
}

Input Parameters

Parameter	Type	Default	Description
exchanges	string[]	[Polymarket, Kalshi, Limitless]	Exchanges to query
query	string	—	Keyword / natural-language search. Omit for top events by volume.
limit	number	50	Max results per exchange (1–1000)
status	string	active	active | inactive | closed | all
category	string	—	Category filter (e.g. politics, crypto, sports)
tags	string[]	—	Tag filter — returns events matching ANY tag
minVolume	number	0	Minimum total traded volume (USD)
minLiquidity	number	0	Minimum market liquidity (USD)
includeOutcomes	boolean	true	Include full outcomes array (for multi-outcome markets)
includeSourceMetadata	boolean	false	Attach raw venue-specific metadata to each event
crossVenueDedup	boolean	false	Merge same-event records from different exchanges
outputFormat	string	events	events | markets | flat (see below)

---

Output Formats

events — One record per event, markets nested

{
  "id": "...",
  "title": "Will Trump nominate Kevin Warsh as Fed Chair?",
  "sourceExchange": "Polymarket",
  "volume": 4820000,
  "volume24h": 130000,
  "category": "politics",
  "tags": ["US", "Fed", "economy"],
  "markets": [
    {
      "marketId": "...",
      "title": "Kevin Warsh",
      "yes": { "price": 0.72, "priceChange24h": -0.03 },
      "no":  { "price": 0.28, "priceChange24h": 0.03 },
      "liquidity": 920000,
      "volume": 4820000,
      "resolutionDate": "2026-01-31T00:00:00Z",
      "status": "active"
    }
  ],
  "scrapedAt": "2026-06-02T10:00:00Z"
}

markets — One record per market, event info flattened

{
  "marketId": "...",
  "eventTitle": "Will Trump nominate Kevin Warsh as Fed Chair?",
  "title": "Kevin Warsh",
  "yes": { "price": 0.72 },
  "no":  { "price": 0.28 },
  "volume": 4820000,
  "liquidity": 920000,
  "resolutionDate": "2026-01-31T00:00:00Z",
  "sourceExchange": "Polymarket",
  "scrapedAt": "2026-06-02T10:00:00Z"
}

flat — One record per outcome (Yes / No / multi-outcome)

Best for price comparison, arbitrage detection, and time-series storage.

{
  "eventTitle": "Will Trump nominate Kevin Warsh as Fed Chair?",
  "marketTitle": "Kevin Warsh",
  "outcomeLabel": "Yes",
  "price": 0.72,
  "priceChange24h": -0.03,
  "volume": 4820000,
  "liquidity": 920000,
  "resolutionDate": "2026-01-31T00:00:00Z",
  "status": "active",
  "sourceExchange": "Polymarket",
  "url": "https://polymarket.com/...",
  "scrapedAt": "2026-06-02T10:00:00Z"
}

All prices are normalized to 0.0–1.0 (implied probability) regardless of venue.

---

Under the Hood

This actor is built on pmxtjs, the TypeScript SDK of the PMXT unified prediction market API — think of it as the ccxt for prediction markets.

Core Data Flow

fetchEvents(params)          ← pmxtjs Exchange class
      │
      ▼
  UnifiedEvent[]             ← normalized, cross-exchange schema
      │
  ┌───┴──────────────────────────┐
  │  filter (volume, liquidity)  │
  │  dedup (crossVenueDedup)     │
  └───────────────┬──────────────┘
                  │
            Apify Dataset
      (events | markets | flat)

Data Model

UnifiedEvent
  ├─ id, title, description, slug
  ├─ url, image, category, tags[]
  ├─ volume, volume24h
  ├─ sourceExchange
  └─ markets: UnifiedMarket[]
        ├─ marketId, title, description
        ├─ resolutionDate, status
        ├─ volume, volume24h, liquidity, openInterest
        ├─ yes: MarketOutcome { price, priceChange24h }
        ├─ no:  MarketOutcome { price, priceChange24h }
        └─ outcomes: MarketOutcome[]   ← multi-outcome markets

fetchEvents Parameters (pmxtjs)

The actor maps its input directly to pmxtjs fetchEvents params:

await api.fetchEvents({
  query:   'Who will be Fed Chair?',  // natural-language search
  limit:   50,                        // max results
  status:  'active',                  // active | inactive | closed
  category: 'politics',
  tags:    ['US', 'Fed'],
  // plus: offset, cursor, sort, eventId, slug, series
});