# Polymarket (/docs/polymarket)
Polymarket [#polymarket]
Search Polymarket prediction markets via the public Gamma API — a server-side multi-query research endpoint that ranks results by topic relevance.
Base URL: `/v1/polymarket/...`
What you get [#what-you-get]
Polymarket events surface real-money prediction-market signals: who's favoured to win the next US election, will Bitcoin hit $200k, will OpenAI ship GPT-5 by year-end. Each event carries:
* Outcome prices (`Yes` / `No` or per-candidate, normalised 0–1)
* 24h / 1w / 1m volume and liquidity in USD
* Day / week / month price changes
* Market end dates
* "Competitiveness" — how close to 50/50 the market is
The research endpoint [#the-research-endpoint]
`GET /v1/polymarket/research` is the whole platform. It takes a free-text `query` and does
the fan-out for you: it expands the topic into up to six search queries (the core subject
plus its individual informative words), issues them in parallel, de-duplicates events by
id, drops matches that share no informative word with your topic, and ranks what is left
by topic-to-title similarity. Framing prefixes like "last 30 days" or "what are people
saying about" are stripped before expansion, so natural-language topics work as written.
`limit` (default 10) caps results per result type on each fan-out call, which is how you
bound the response size.
`/v1/polymarket/search`
, the raw single-query lookup, was withdrawn on 2026-06-06
when the Gamma API began returning
`502`
on consecutive probes. It is disabled and returns
a
`503`
at no charge.
`research`
covers the same ground with multi-query expansion, so
there is nothing to migrate beyond the path.
Getting started [#getting-started]
```bash
curl "https://www.socialcrawl.dev/v1/polymarket/research?query=bitcoin%20halving&limit=20" \
-H "x-api-key: sc_your_api_key_here"
```
Response shape [#response-shape]
The research endpoint normalises to the unified `SearchResult` archetype:
```json
{
"success": true,
"platform": "polymarket",
"endpoint": "/v1/polymarket/research",
"data": {
"items": [
{
"id": "evt-...",
"title": "Will Trump win the 2028 election?",
"slug": "trump-2028",
"markets": [
{
"question": "Will Trump win the 2028 election?",
"outcomes": "[\"Yes\",\"No\"]",
"outcomePrices": "[\"0.32\",\"0.68\"]",
"liquidity": 10000,
"volume": 50000,
"endDate": "2028-11-07T00:00:00Z",
"oneDayPriceChange": 0.01,
"oneWeekPriceChange": -0.02,
"oneMonthPriceChange": 0.05
}
],
"tags": [{ "label": "Politics" }]
}
]
},
"credits_used": 5,
"credits_remaining": 95,
"request_id": "req-...",
"cached": false
}
```
Polymarket is a passthrough — fields under `data.items[]` are exactly what the Gamma API returns. There is **no** field map and **no** computed-field layer (no `engagement_rate`, `language`, etc.) because prediction-market events don't have authors or text content the way a TikTok post does.
Notes [#notes]
* `GET` with query parameters.
* Authentication via `x-api-key` on the SocialCrawl side. Polymarket's upstream is no-auth, but you still authenticate to us.
* Responses cache for 2 minutes. Markets move fast, so send `Cache-Control: no-cache` when you need a billed fresh read.
* The `research` endpoint costs **5 credits** because one request makes six or more upstream calls (multi-query expansion, parallel fan-out, then ranking). That is the price whichever `limit` you pass.
* There is no pagination: the upstream returns a single non-cursored result set, and `limit` is the only size control.