What the interval is
Perpetual futures have no expiry, so exchanges keep their price tethered to spot with funding: a periodic payment between longs and shorts. The funding INTERVAL is how often that payment settles — classically every 8 hours, but increasingly every 4 hours, and on some venues and symbols every hour.
Only position holders at the settlement moment pay or receive. Between settlements the predicted rate drifts with the premium between perp and spot price — which is why a rate you saw an hour ago is not the rate you will be paid.
The map across exchanges — and within them
Intervals are not a per-exchange constant — they vary per SYMBOL within the same venue. Majors like BTC and ETH usually settle every 8 hours; volatile new listings often start at 4 hours or 1 hour so the tether holds; some venues adjust intervals dynamically when a contract trades far from spot.
This is a live attribute of every instrument, not something to memorize from a docs page. Arbitron's funding pages show the actual interval per pair next to every rate (the Interval column), fetched from exchange data rather than assumed.
The APR math: same rate, 8× difference
A funding rate is paid PER SETTLEMENT, so annualizing depends entirely on the interval: 0.01% every 8 hours is 3 settlements a day ≈ 11% APR, while 0.01% every hour is 24 settlements ≈ 88% APR. Two rows showing the same "0.01%" can differ 8× in actual yield.
Funding arbitrage adds a second trap: a spread between a 1h-interval leg and an 8h-interval leg does not settle symmetrically — one side pays eight times while the other pays once. Comparing raw rates across different intervals without normalizing is the single most common mistake in DIY funding-arb spreadsheets.
Why hardcoded assumptions fail
Tools built when "funding = every 8 hours" was universally true silently mis-annualize everything on modern 4h/1h symbols — overstating or understating APR by 2–8×. Worse, an interval can change for an existing symbol after a volatility episode, breaking yesterday's correct assumption.
Arbitron never hardcodes intervals: connectors read them from each exchange's API or infer them from consecutive settlement timestamps, per symbol, continuously. When you compare funding opportunities in the scanner, the normalization is already done — but if you build your own sheets, normalize per-interval first or the best-looking rows will be the wrong ones.