It is the line traders notice last and complain about most. It is not a fee, it is not charged by the venue, and on a position held for weeks it can cost more than every trading fee you paid to open and close it.
A traditional futures contract has an expiry date. On that date it settles against the spot price, and the certainty of that settlement is what keeps the two prices tethered in the meantime: if the future drifts too far from spot, somebody can arbitrage the gap knowing exactly when it closes.
A perpetual contract has no expiry. It never settles, so that mechanism does not exist, and nothing structural stops it from drifting away from the spot price and staying there. Funding is the replacement. It is a payment made directly between the traders holding the contract, at fixed intervals, in the direction that pushes the contract price back towards spot.
This is the single most important thing to understand about it: the exchange does not collect funding. It moves it from one side of the book to the other, and takes nothing.
When the perpetual trades above spot, longs pay shorts. When it trades below spot, shorts pay longs. The payment happens at fixed times, on most venues every eight hours, and it is calculated on the notional value of the position rather than on the margin backing it.
The logic is a feedback loop. If the contract is expensive relative to spot, holding a long becomes expensive too, which discourages new longs and encourages shorts, which pushes the price back down. If the contract is cheap, the reverse. The rate does not force the price anywhere; it makes one side of the trade progressively costly until the imbalance corrects itself.
Funding rates are usually quoted per interval, and that is where the arithmetic misleads. A rate that looks negligible over eight hours is paid three times a day, and a position held for a month sees roughly ninety of those payments. The number that matters is never the one displayed next to the contract; it is that number multiplied by how long you intend to stay.
It compounds against a specific kind of trader: the one who holds a directional position for weeks because the thesis is a slow one. Someone opening and closing within a session may never pay funding at all. Someone holding a leveraged long through a sustained bull phase, when almost everybody is long and the perpetual sits above spot, pays it continuously and in the worst possible direction.
The pattern to remember: funding is usually most expensive exactly when the crowd agrees with you, because that is precisely when the contract trades furthest from spot.
Every venue publishes the current rate and, on most, the predicted rate for the next interval. Two habits make the difference between a cost you accepted and a cost you discovered.
A positive rate means longs pay. A negative rate means shorts pay. Entering a long into a strongly positive rate means you are paying for the privilege from the first interval, and the size of the rate tells you how crowded that side already is. It is one of the few honest sentiment indicators available, because it is backed by actual money changing hands rather than by a survey.
Before opening, take the current rate, multiply it by the number of intervals you expect to hold, and compare that number to the move you are actually expecting. If funding over your horizon eats a meaningful fraction of your target, the trade needs to be re-sized or the horizon shortened. This calculation takes ten seconds and almost nobody does it.
Trading fees are charged by the venue, published in a schedule, and paid on every fill. Funding is charged by the other side of the book, varies continuously, and is paid only if you are holding at the wrong moment. Confusing the two leads to two different mistakes, in opposite directions.
The first is blaming the exchange for funding costs, which is like blaming the auctioneer for the price. The second, and the more expensive one, is comparing two venues on their published fee schedules while ignoring that a persistent funding differential between them can dwarf the fee difference entirely over any holding period longer than a day.
It is also why a fee rebate and a funding cost are not comparable objects. A rebate returns part of what the venue charged you. Funding never went to the venue in the first place, so nothing about it can be rebated by anybody. Any offer suggesting otherwise is describing something else.
Funding exists on perpetual contracts, and perpetual contracts are a crypto invention. They were introduced because crypto had no established futures infrastructure, no clearing houses willing to touch the asset class, and a retail audience that wanted leveraged exposure without ever taking delivery of anything. The contract that answered all three has no expiry, which is precisely why it needs funding to stay honest.
That origin has a consequence worth understanding. Because the perpetual is where most crypto leverage lives, the funding rate is one of the cleanest readings available of how leveraged the market currently is, and in which direction. It is not a forecast and treating it as one is expensive. But it is a measurement, taken continuously, of something that no survey and no social media sentiment index can measure: money actually committed.
The patterns are consistent enough to be worth naming. Rates that stay strongly positive for weeks describe a market where leveraged longs dominate and are paying continuously to keep that exposure. Rates that flip sharply negative usually accompany a violent decline, because the traders who were paying have been liquidated and the ones who remain are being paid to hold a short. A rate that hovers near zero on a large pair usually means neither side is crowded, which is the least interesting reading and the most common one.
None of this is a reason to enter a trade, and the number of people who have lost money selling a crowded long or buying a crowded short is not small. What it is good for is calibrating cost. A market where the rate has been strongly positive for a month is a market where a leveraged long is structurally expensive to hold, and that belongs in the sizing decision whether or not the direction turns out to be right.
Funding is the only sentiment indicator in crypto that is backed entirely by money changing hands. That makes it honest, and it does not make it predictive.
Three levers, and only three, in order of how much they move the number.
Funding scales with time held and nothing else. A thesis that needs three weeks to play out will pay three weeks of funding, and there is no clever way around that. The honest response is to include it in the expected cost of the trade before entering, not to discover it afterwards.
When the rate is strongly positive, shorts are being paid to hold. This does not make a short a good trade, and treating funding as a reason to take a direction is one of the fastest ways to lose more than you collect. But when the analysis already points one way, a favourable rate is a genuine tailwind and an unfavourable one is a genuine headwind, and both belong in the decision.
If a position is going to be closed within hours anyway, closing it before the funding timestamp rather than after costs nothing and saves the payment entirely. This is not a strategy, it is housekeeping, and over a year of active trading it is not nothing.
The rate is not decided by anybody at the venue. It is computed, every interval, from two components. The first is the premium: how far the perpetual traded from an index of spot prices over the interval, averaged rather than sampled at a single instant. The second is a small fixed interest component, meant to reflect the cost of holding the two currencies in the pair, and it is rarely the part that decides anything.
The premium does almost all the work. When the contract spent the interval above the index, the premium is positive and the rate follows it up; when it spent the interval below, the reverse. Most venues then clamp the result inside a band, so that a violent dislocation cannot produce an absurd payment, and a few publish the two components separately, which means the number can be checked rather than trusted.
This is worth knowing for one practical reason. Because the premium is averaged over the whole interval, a sharp move in the last minutes before a timestamp barely shifts the payment that is about to be taken. Traders who watch the rate tick up during a rally and expect it to be reflected immediately are reading a number that is mostly already decided.
No. Funding moves between traders holding the contract: longs to shorts when the rate is positive, shorts to longs when it is negative. The venue calculates and settles it but takes no part of it. This is different from a trading fee, which the venue does collect and publishes in its schedule.
On most venues every eight hours, at fixed times of day. You pay or receive only if your position is open at that exact moment. A position opened and closed between two timestamps pays no funding at all, no matter how large it was or how much it moved.
On the notional value of the position. A position of one hundred thousand notional pays the same funding whether you opened it with ten thousand of margin at ten times leverage or with fifty thousand at two times. Leverage changes your liquidation distance, not your funding bill.
Easily, on anything held longer than a day or two. Trading fees are paid twice, once to open and once to close. Funding is paid every interval for as long as you hold. Over a month, a position on the paying side of a persistent rate can cost several times what its entry and exit cost combined.
No. Eight hours is the most common, but some venues settle every four hours, some every hour, and a few adjust the interval during extreme volatility. Comparing two rates without checking their intervals is comparing two different numbers, and the mistake always flatters the venue with the shorter interval.
No. Funding exists only on perpetual contracts, because it exists to solve a problem perpetuals have and spot does not: keeping a contract with no expiry tied to the price of the thing it tracks. Buying an asset outright costs you the trading fee and nothing recurring.
Because leveraged demand in crypto skews long. More traders want leveraged upside exposure than leveraged downside exposure, which pushes the perpetual above spot most of the time, which makes the rate positive most of the time. The sustained negative rates tend to appear in sharp declines, when the crowd flips.