Rate limits and data freshness

Know the limits for each token and how far market data can lag behind the source.

Rate limits and quotas

Each API key has a per-minute request limit. Going over returns 429 with a Retry-After header suggesting how many seconds to wait.
Some plans have a daily request limit, counted by UTC date and reset at 00:00 UTC.
Every response includes X-RateLimit-Limit (the per-minute limit) and X-RateLimit-Remaining (requests left).
The WebSocket handshake also counts as a request.

Data freshness

For stability and rate limiting, market data endpoints are briefly cached, so data can lag the source slightly. Below are the default cache times; the X-Proxy-TTL header on each response gives the actual value in seconds.

Data	Default cache time
Real-time quotes, ticks, order book	3 seconds
Minute candles (including the current one)	30 seconds
Day, week and month candles (including the current one)	5 minutes
Historical candles with the `et` parameter	1 hour
Symbol list, holidays, basic info	24 hours
Market status	30 seconds

Cache response headers

X-Proxy-Cache	Meaning
`HIT` / `L2-HIT`	Served from cache
`MISS`	Fetched from the upstream for this request
`COALESCED`	Merged with an identical concurrent request that shared one upstream fetch
`STALE`	The upstream had a brief problem, so data up to 30 seconds old was returned
`BUSY`	The upstream call budget is temporarily used up; retry shortly (returns 503)
`BYPASS`	This request is not cached

For high-frequency trading that needs millisecond latency, connect to the data source or exchange directly instead of relying on cached market data.