Funding looks tiny per interval and compounds into a real cost. Convert any funding rate into an annualized figure and see what holding your position actually costs per day — the number that quietly decides whether a slow trade can ever pay.
Positive funding: longs pay shorts. Negative: shorts pay longs. Rates reset every interval, so annualized figures assume the current rate persists — in practice it mean-reverts, which is exactly why extreme funding is more often a positioning signal than a durable yield.
APR = rate per interval × intervals per year. A 0.01% rate every 8 hours is 3 payments a day, 1,095 a year — 10.95% annualized. Holding cost = notional × rate × payments per day. On a $10,000 position that "tiny" 0.01% is $3 a day, $90 a month.
Positive funding means longs pay shorts; negative means shorts pay longs. The rate resets every interval and mean-reverts hard, so annualized numbers assume persistence that rarely holds — which is why extreme funding readings work better as a crowd-positioning signal than as yield.
When the funding rate is positive, longs pay shorts; when negative, shorts pay longs. The mechanism exists to pull the perpetual's price toward the spot index: whichever side is more crowded pays the other side to take the opposite exposure.
Sometimes — extreme positive funding means longs are crowded and paying heavily to stay in, which historically precedes mean reversion more often than continuation. But 'historically precedes' is a hypothesis to test, not an edge to assume: fees and timing eat most naive funding-fade strategies. Measure it on out-of-sample data before trading it.
Barely — a scalp that's flat before the next funding timestamp pays nothing. It matters enormously for anything held days or weeks: at 20%+ annualized (common in bull markets), funding alone can exceed the entire expected edge of a slow strategy. Always compare your expected move against the carry cost over your expected holding time.
Carry cost is one of the silent killers of otherwise-decent strategies. Find out if yours clears all of its costs — graded honestly on out-of-sample data.
Grade your strategy — free