Method
Every number on this site comes from probes this server ran itself. Nothing is aggregated from an exchange's own status page, and nothing is estimated.
Explanation updated: 2026-09-10 · About Tickmark
What these figures can tell you
We observe public exchange APIs and funding rates reported by venues. API performance does not establish the safety of funds or guarantee orders, cancellations or withdrawals. Funding rates describe a holding cost, not a fixed return.
How reliability is measured
Public time, market-data and order-book interfaces are checked approximately every 15 seconds; a request unfinished after 8 seconds times out. Availability measures the share of valid responses. Latency includes the network path from the observation location to the venue.
Read the location, observation period and sample count together. p95 and p99 mean that 95% and 99% of observed requests completed within the indicated time. A short record from one location does not represent every region or long-term performance.
Failures and missing observations
An incident requires 2 consecutive failures; shorter disruptions may not appear in the incident list. Recognised regional access restrictions are separated from service failures. Missing data is labelled, not counted as working service or a zero fee.
Clock offset and market-data age are estimates made through network requests. Small differences can reflect network variation or timing uncertainty and should not be treated as conclusive advantages.
Availability when the market moves
Uptime averaged over a calm market says little about the minutes that decide whether a position can be closed. Availability is therefore also reported inside the busiest 90th-percentile 5-minute windows in the record, measured on the consensus Bitcoin price — the median across venues, so one venue's bad tick cannot create a window. That currently means a move of at least 0.18%.
The threshold comes from the record rather than being fixed. A fixed one ran here for half a day without firing once, because the largest five-minute move in that period was smaller than it. A floor still applies, so a genuinely flat record reports no windows instead of presenting its calmest minutes as a stress test. Read the threshold with the figure: it says what «moving» meant for this record.
How funding rates are compared
Match the contract and settlement currency before comparing the rate and interval. Current comparisons require a known observation time no more than 30 minutes before the page update and sufficiently close observation times. Unconfirmed intervals are shown separately.
24-hour equivalent = rate per settlement × 24 ÷ interval in hours. Multiply by 365 for the annualised equivalent. This assumes an unchanged rate, not fees already paid or a future return. Differences between percentages use percentage points. Positive funding means longs pay shorts; negative funding reverses the direction.
What a cost estimate includes
The amount entered is position value, not margin. Funding per settlement = position value × rate. Payment or receipt depends on position direction and whether the position is held at settlement. Trading fees, bid–ask spread, slippage and price changes are separate.
Order-book price costs are estimated from visible orders. Orders can change before execution, so the estimate does not guarantee a fill price and excludes trading fees. When depth is insufficient, a partial fill must not be quoted as the cost of the whole order.
History and update times
Recorded observations retain their actual observation time for each venue and contract, with at most one record per hour shown. They are observed rates, not official settlement statements. Gaps are not filled by assuming unchanged rates. Legacy records without reliable observation times are not shown as settled history.
Reliability, funding and order-book data carry separate times. Automated data updates differ from revisions to explanatory text; check the time of the data you are using. Our record begins at 2026-09-11 00:50 UTC.
Sources, responsibility and corrections
The footer links to sources, site information and contact details. Report discrepancies with the page, venue, contract, time and an official source or screenshot to [email protected]. Do not send passwords, keys or account statements. Referral relationships must not change measurements.