How we count
The same figures mean different things on different venues. Below are the rules we use to bring them to one basis, and what follows from that for the numbers on this site.
Open interest is reduced to the one-sided figure
Some exchanges publish open interest two-sided, counting both the long and the short leg of the same contract.
The two-sided figure is exactly twice the one-sided one: every open position has a counterparty. Taken as-is, a venue reporting that way looks twice as large as its neighbours, and any screener comparing venues starts systematically singling it out. We take the one-sided figure wherever the exchange provides it, and halve the two-sided one where it does not.
Consequence: on some venues our open interest is exactly half of what you see where the format is not normalised. That is normalisation, not under-reporting.
Funding rates are normalised to an eight-hour basis
A funding rate is the payment FOR ONE settlement interval, and that interval stopped being uniform a while ago.
Four-hour settlement is now at least as common as eight-hour, and on some venues funding is charged hourly. A 0.01% rate on an hourly interval and the same 0.01% on an eight-hour one differ eightfold in cost, and comparing them directly is meaningless. We track the interval per pair and normalise every rate to a common eight-hour basis before any comparison.
Consequence: our rate for a pair settling hourly or every four hours will differ from the raw number on the exchange. The raw number is right for its own venue and wrong for comparison.
Coins and instruments are counted separately
One coin traded on ten venues is one coin and ten instruments.
Every minute the screener re-evaluates instruments: each pair, on each venue, on each market, separately. That is why the coverage figures on this site show both numbers side by side and never substitute one for the other — they answer different questions. A spot pair and a futures pair for the same coin on the same exchange are two different instruments too: different prices, different volumes, different behaviour.
Consequence: our instrument count is several times the coin count, and the smaller of the two does not describe how much work the screener does.
Only dollar-quoted pairs enter the analysis
Turnover of a pair quoted in lira or in bitcoin is not dollars, and cannot be compared against a dollar threshold.
Liquidity thresholds are set in dollars, and volume times price yields dollars only for pairs quoted in the dollar and its equivalents. The error here is asymmetric and therefore especially damaging: a pair in a national currency would overstate turnover many times over and slip through a threshold it does not meet, while a pair quoted in bitcoin would understate it by orders of magnitude and never fire at all — silently, without a trace.
Consequence: pairs on non-dollar quotes do not take part in the screeners. That is a deliberate narrowing, not an oversight.
You do not have to take our word for it
The speed and coverage figures on this site come from the running service, not from the copy.
The tiles showing market coverage, sweep speed and daily workload are read from the live screener on every page render. When the service is unavailable or still warming up after an update, the tile disappears entirely — it never shows a zero and never shows yesterday's value. That is deliberate: a constant in the copy quietly becomes untrue one day, while a missing block is obvious at once.