Skip to main content
The Futuur API enforces rate limits per authenticated user to ensure fair access and platform stability. Several limits can apply to one request. The request is rejected as soon as any of them is exceeded.

Limits

All HMAC requests

Both limits apply to your account, not to an IP address or a single key: every HMAC key of the account shares one bucket. They cover all endpoints together — different paths and querystrings count against the same bucket.

Per-endpoint limits

These apply on top of the account-wide limits.
Order creation is the tight limit. You can send 30 create requests per minute, not 2,000. POST /orders/signing-context/, cancels, and reads do not count against the 30. Size a quoting loop around that number.
The per-endpoint limits use a rolling 60-second window. The account-wide limits reset on the UTC minute and the UTC day.

What happens when you exceed a limit

When you exceed a rate limit, the API returns HTTP 429 Too Many Requests with code throttled and a Retry-After header giving the seconds to wait. Back off until the indicated time.
429 Too Many Requests
Requests carrying HMAC headers that fail authentication are also limited per IP. After 300 failures in a minute, further HMAC requests from that address get 429 until the minute ends.

Retrying a POST

POST /orders/, POST /orders/batch/, and POST /orders/batch-update/ accept an optional Idempotency-Key header. If you repeat the same POST with the same key and the same body after a successful response, the API returns the stored response and does not create another order. A repeat sent while the first request is still running returns 409 with code idempotency_key_in_progress. A different body is a different request, so requoting at a new price is not blocked.
On the phase-one preview host a signed order can only be accepted once: the first create consumes its nonce. A retry after a timeout returns order_intent_nonce_stale when the first attempt went through. List your orders before you sign again.

Best practices

Use the include_mine parameter wisely. Order book and event detail reads are served from cache. By default they return the public payload; pass include_mine=1 only on the calls where you actually need your own orders or wagers embedded, so the rest of your polling stays on the fast path. Prefer polling intervals of 1 second or more. Responses are cached for a few seconds server-side, so faster polling returns the same data while consuming your quota. Cache responses where possible. Market data like event listings and price history does not change every second. Cache responses and serve them from memory rather than re-fetching on every operation. Use exponential backoff on 429 responses. If you receive a 429, wait before retrying. Start with a short delay (e.g., 1 second) and double it on each successive failure, up to a maximum wait time.
python
HMAC headers carry a timestamp, so build fresh headers for every retry instead of reusing the ones from the first attempt.