Rate limits
Each key has three budgets, counted over one minute:
Cancellations count against no budget:
POST /v1/orders/{id}/cancel and POST /v1/orders/cancel-all are not rate-limited per key, so you can always pull your orders.
The budgets are separate: reading orders does not consume the placement budget. Over a budget, requests get 429 RATE_LIMITED with these headers:
Wait
retry-after seconds before you retry. To place many orders, spread them over the minute rather than bursting.
Health check
GET /v1/health carries no key, so it has its own budget per address: 120 requests per minute. Over it, requests get the same 429 RATE_LIMITED and headers. Its result is refreshed at most every 15 seconds, so polling every 15 seconds is enough.
Failed authentication
After 20 failed authentications from the same address within one minute, every request from that address gets429 RATE_LIMITED until the minute ends, whatever key it carries. A client that retries with a revoked key can lock itself out this way: check the key first. See Authentication.
Request limits
Your account also has its own limits, such as a maximum stake per fixture. They are applied to every order and shown by
GET /v1/betslip.
WebSocket
Over the connection, fixture or selection limit, the server refuses with
4004 limit_exceeded. See Close codes for what each code means and what to do.