One feed for Polymarket, Kalshi and Manifold

Quick start

One event, with its markets and the matched question on the other venue:

curl -s https://predictions.io/api/v1/events/EVENT_ID \
  -H "X-API-Key: YOUR_KEY"

curl -s https://predictions.io/api/v1/events/EVENT_ID/matches \
  -H "X-API-Key: YOUR_KEY"

Every response uses the same envelope, so a client only needs one parser:

{
  "ok": true,
  "data": {
    "id": "…",
    "platform": "polymarket",
    "title": "Will the Fed cut rates in September?",
    "volume": 4215330,
    "markets": [
      { "id": "…", "title": "Yes", "volume": 4215330,
        "outcomes": [ { "name": "Yes", "price": 0.62 },
                      { "name": "No",  "price": 0.38 } ] }
    ]
  },
  "meta": { "request_id": "req_…", "timestamp": "…", "cached": false }
}

Endpoints

All read-only, all under /api/v1. Cache times are how long a repeated call may return the same payload — the meta.cached flag tells you when it did.

Worked examples

What the main calls return once you have a key. Responses are trimmed for length; field names and shapes are exact.

1. The same question on another venue

The core of the feed: given one event, which events on the other venues have been validated as the same question. Pair this with /events/:id on each side and you have both venues' prices for one bet.

GET /api/v1/events/EVENT_ID/matches
X-API-Key: YOUR_KEY

{
  "ok": true,
  "data": [
    {
      "matched_event_id": "KXFEDDECISION-26SEP",
      "matched_event_title": "Fed decision in September?",
      "matched_platform": "kalshi",
      "similarity": 0.91,
      "confidence": 96
    }
  ],
  "meta": { "request_id": "req_…", "timestamp": "…", "cached": true, "cache_age_s": 41 }
}

2. Browse by topic

Lobbies are the human-curated topic hubs on the site. List them, then pull one with its approved events and markets.

GET /api/v1/lobbies
X-API-Key: YOUR_KEY

{
  "ok": true,
  "data": [
    {
      "id": 6,
      "title": "US Politics",
      "description": "Elections, Congress, the White House…",
      "event_count": 142,
      "review_count": 0,
      "images": { "description": { "avif": "…", "webp": "…" },
                  "background": { "avif": "…", "webp": "…" },
                  "sidemenu":   { "avif": "…", "webp": "…" } }
    }
  ],
  "meta": { "request_id": "req_…", "timestamp": "…", "cached": false }
}

GET /api/v1/lobbies/6            → the lobby with events, markets, volume + sentiment stats
GET /api/v1/lobbies/6/summary    → its AI-written market summary

3. Price history for a chart

Points are t (Unix seconds) and p (probability 0–1), downsampled to your fidelity in minutes. Ranges: 1d, 7d, 30d, 90d, all. The window anchors on the last data point, so a settled market still returns its final days.

GET /api/v1/markets/MARKET_ID/price-history?range=7d&fidelity=60
X-API-Key: YOUR_KEY

{
  "ok": true,
  "data": {
    "market_id": "MARKET_ID",
    "platform": "polymarket",
    "points": [
      { "t": 1756771200, "p": 0.58 },
      { "t": 1756774800, "p": 0.60 },
      { "t": 1756778400, "p": 0.62 }
    ],
    "updated_at": "2026-09-03T18:20:11.000Z"
  },
  "meta": { "request_id": "req_…", "timestamp": "…", "cached": false }
}

4. In code

const res = await fetch(
  "https://predictions.io/api/v1/events/EVENT_ID",
  { headers: { "X-API-Key": process.env.PREDICTIONS_KEY } }
);
const { ok, data, error } = await res.json();
if (!ok) throw new Error(error.code + ": " + error.message);

for (const m of data.markets) {
  const yes = m.outcomes.find((o) => o.name === "Yes");
  console.log(m.title, yes?.price);
}

import os, requests

r = requests.get(
    "https://predictions.io/api/v1/events/EVENT_ID/matches",
    headers={"X-API-Key": os.environ["PREDICTIONS_KEY"]},
    timeout=10,
)
body = r.json()
if not body["ok"]:
    raise RuntimeError(body["error"]["code"])

for m in body["data"]:
    print(m["matched_platform"], m["matched_event_title"],
          m["confidence"])

5. Errors, limits and caching

Errors use the same envelope with ok: false. Codes: BAD_REQUEST 400, UNAUTHORIZED 401, FORBIDDEN 403, NOT_FOUND 404, RATE_LIMITED 429, INTERNAL_ERROR 500, SERVICE_UNAVAILABLE 503.

HTTP/1.1 429 Too Many Requests
Retry-After: 23
X-RateLimit-Limit: 100
X-RateLimit-Remaining: 0
X-RateLimit-Reset: 1756778460

{
  "ok": false,
  "error": { "code": "RATE_LIMITED",
             "message": "Rate limit exceeded",
             "retry_after_s": 23 },
  "meta": { "request_id": "req_…", "timestamp": "…" }
}

Every successful response carries the three X-RateLimit-* headers and an ETag. Send it back as If-None-Match and an unchanged payload returns 304 without counting against your body transfer; meta.cached and meta.cache_age_s tell you how fresh a cached hit is.

What the unified layer adds

The venues' own APIs are good. What they can't do is agree with each other — three price formats, three volume definitions, three ways to say "this question". The feed does that work once:

Polymarket API vs Kalshi API vs Manifold API

If you want to go straight to the source, here is how the three venues' own APIs differ — and why normalizing them is most of the work.

Questions developers ask