Crypto Perpetual Funding Rates Explained: Formulas, Payments and Trading Costs

Editorial responsibility: Oliver Sinz · Last verified: October 14, 2026

A funding rate is the periodic payment that passes between the long and short sides of a perpetual futures contract. When the rate is positive, longs pay shorts. When it is negative, shorts pay longs. The payment is a percentage of the position's notional value, not of the margin posted, so a rate that looks tiny per interval can add up to a meaningful cost or receipt over a holding period.

This guide explains how venues construct the rate, how a single payment is calculated, how payments accumulate, how to annualize them without misleading yourself, how inverse (coin-settled) contracts change the currency of the payment, and how funding debits can push a leveraged position toward liquidation. Every price, rate and return in the examples is a hypothetical illustration, not market data or a forecast. Venue rules were checked against official documentation on October 14, 2026.

Why perpetual futures need funding

A dated futures contract is pulled toward the spot price because it settles at expiry. A perpetual never settles, so nothing forces its price back to spot. Funding replaces that anchor with an incentive: when the perpetual trades above its spot-based index, holding a long becomes costly and holding a short pays, which rewards trades that narrow the gap. The reverse applies when the perpetual trades at a discount.

Two points are easy to miss. First, funding is a transfer between traders, not exchange revenue; Binance, for example, states that it does not charge fees on funding payments. Second, funding is an incentive, not a guarantee. In their study of perpetual pricing, He, Manela, Ross and von Wachter find that deviations of crypto perpetuals from no-arbitrage prices are larger than in traditional currency markets, although those deviations have diminished over time.

How the funding rate is built

Binance, Bybit, OKX and Hyperliquid all combine two ingredients:

• An interest rate component (I), a fixed parameter meant to reflect the difference in borrowing costs between the quote currency and the base asset. Binance, Bybit and OKX each document a default equivalent to 0.01% per 8 hours (0.03% per day) for standard contracts, with exceptions.
• A premium index (P), which measures how far the perpetual trades from its spot index. Binance and Bybit build it from impact prices, the average fill price of a standard-size market order against the order book:

Premium index = [max(0, Impact bid − Index) − max(0, Index − Impact ask)] / Index

The samples are averaged over the funding interval. Binance samples every five seconds and gives later samples more weight on intervals longer than an hour; Bybit samples every minute, and its 8-hour example likewise weights later minutes more heavily. That weighting matters for interpretation: the rate a venue displays midway through an interval is a provisional estimate, not the rate that will settle.

The two ingredients are then combined with a clamp, sometimes called a damper:

F = P + clamp(I − P, −0.05%, +0.05%)

The clamp limits how far the interest term can pull the rate away from the premium. It has a revealing side effect. With I = 0.01%, any average premium between −0.04% and +0.06% produces a funding rate of exactly 0.01%:

That band explains why a rate of exactly 0.01% per 8 hours can recur for long stretches on venues using this structure, even while the perpetual trades slightly below its index. Outside the band, the rate equals the premium adjusted by 0.05 percentage points toward the interest rate.

Venues add further rules. Binance and OKX divide the result by 8/N, where N is the interval in hours, so a contract settling every four hours pays half the 8-hour rate per settlement. Bybit's documentation instead scales the interest component to the interval length. All three cap and floor the final rate: Binance ties the cap for a published list of major symbols, such as BTCUSDT, to 0.75 times the maintenance margin ratio and uses ±2% for other USDⓈ-M contracts, while Bybit derives its limits from each contract's lowest risk-limit tier. All three can also shorten the settlement interval, down to hourly, when the rate hits its limit.

Venue conventions are not interchangeable

The structure above is common but not universal. The table summarizes how five venues documented funding when this guide was verified.

Three practical consequences follow. On snapshot venues, only positions open at the funding time pay or receive, so a position closed beforehand pays nothing for that interval, although Binance notes a 15-second deviation in actual funding transaction times and Bybit does not guarantee how trades within five seconds of the timestamp are treated. On continuous-accrual venues there is no timestamp to avoid; you pay or receive in proportion to time held. And rates from different venues can only be compared after converting them to the same time basis.

Conventions also change. Coinbase International Exchange, which charged funding hourly, moved its perpetuals to Deribit on October 1, 2026, where funding accrues continuously and settles daily. Check the current contract specification before relying on any rule described here.

Calculating a single funding payment

For a linear contract, such as a USDT-margined BTC perpetual on a snapshot venue, the payment at one funding timestamp is:

Notional        = Position size (coins) × Reference price at the funding time
Funding payment = Notional × Funding rate
Cash flow to a long  = −Funding payment
Cash flow to a short = +Funding payment

The reference price is the mark price on Binance, Bybit and OKX and the oracle price on Hyperliquid. A negative rate flips the direction of the cash flow automatically.

Example 1: one interval, four cases (hypothetical). A trader holds 0.5 BTC on a linear BTCUSDT perpetual. At the funding timestamp the mark price is 64,000 USDT, so the notional is 0.5 × 64,000 = 32,000 USDT.

Two familiar inputs do not appear in the calculation. The entry price is irrelevant: had the trader bought at 60,000, the notional at the timestamp would still be 32,000 USDT. Leverage is irrelevant too: the same 0.5 BTC position pays 3.20 USDT whether 32,000 or 1,600 USDT of margin backs it. Leverage changes what the payment means relative to the trader's capital, and that is where most of the risk lies.

Funding over a holding period

Funding applies at every timestamp you hold through, on the notional at that moment, at whatever rate prevails. Because rates change from interval to interval, the realized cost of a hold depends on the path.

Example 2: a three-day long with variable rates (hypothetical). The trader buys 0.5 BTC at 64,000 USDT, holds through nine 8-hour funding timestamps, and sells at 64,400 USDT. Assume a 0.05% taker fee on each trade. The rates and mark prices below are assumptions chosen to show a period in which long demand briefly rises.

A trader who assumed a constant 0.01% at 64,000 would have estimated 9 × 3.20 = 28.80 USDT. The assumed path costs 65% more, and the largest payment arrives just after the price peak, while the long's notional is still near its high.

The position's result separates into components that should not be blended:

Funding consumed almost a quarter of the price gain. The break-even exit price X, at which the net result is zero, solves:

0.5 × (X − 64,000) − 0.0005 × 32,000 − 0.0005 × 0.5 × X − 47.55 = 0
X = (64,000 × 1.0005 + 47.55 / 0.5) / 0.9995 ≈ 64,159.18 USDT

Without funding, the break-even would be about 64,064.03 USDT. A short over the same path would have lost 200.00 on price and 32.10 in fees while receiving 47.55 in funding, for a net loss of 184.55 USDT. Funding received softens a losing short; it does not rescue it.

Annualizing funding rates

Funding rates are often converted to annual percentages for comparison. The arithmetic is simple, but each conversion rests on assumptions that rarely hold.

At 0.01% per 8 hours and three settlements a day:

• Simple annualization: 0.01% × 3 × 365 = 10.95% of notional per year.
• Compounded: (1.0001)^1,095 − 1 ≈ 11.57%, which assumes every payment is reinvested and the position grows with it.
• Relative to margin: at 5x leverage, 10.95% of notional equals 54.75% of the margin posted.

To compare venues with different intervals, convert every rate to the same basis first. Hyperliquid's hourly payment at an 8-hour rate of 0.01% is 0.00125%, which also annualizes to 10.95%. A contract paying 0.002% every hour, by contrast, is equivalent to 0.016% per 8 hours and annualizes to 17.52% simple, even though its per-settlement figure looks five times smaller than 0.01%.

An annualized figure describes what a rate would add up to if it stayed constant and the position stayed the same size for a year. It is not a forecast, and it is not a return anyone can lock in. The average rate in Example 2 was about 0.0161% per 8 hours, which "annualizes" to roughly 17.64%, yet three days of observations say nothing reliable about the next 362. On snapshot venues the next rate is not settled until its interval ends, and caps, interval changes and shifts in positioning can all move it.

Linear vs. inverse contracts: what currency funding is paid in

On linear contracts, funding is paid in the quote or settlement currency, such as USDT, USDC or USD. On inverse, coin-settled contracts, the position is sized in US dollars, but margin, P&L and funding are all in the underlying coin. Bybit and OKX both compute the coin value of such a position by dividing its dollar size by the mark price:

Position value (BTC)  = Contracts (USD face) / Mark price
Funding payment (BTC) = Position value × Funding rate

Example 3: inverse funding denomination (hypothetical). A trader is long 32,000 contracts of a BTCUSD inverse perpetual with a face value of 1 USD each.

The dollar value of each payment stays constant because the position is a fixed dollar amount; only the number of coins changes. A linear position of a fixed 0.5 BTC behaves the other way: at 80,000 it would pay 4.00 USDT, because its dollar notional rises with the price.

The denomination also changes what a funding receipt is worth. A short collecting funding on an inverse contract accumulates BTC, whose dollar value then moves with the market. That is a collateral-value effect, separate from both funding and price P&L, and it is one reason inverse positions can carry exposure that is not obvious from the contract size. Deribit's BTC perpetual is also inverse, at 10 USD per contract, margined and settled in BTC, but it prorates funding by time: its funding documentation shows a 1 BTC long paying about 0.0000010417 BTC for one minute at a 0.05% 8-hour rate, or 0.0005 × 1 ÷ 480.

Funding, margin and liquidation risk

A funding debit is a realized loss. It reduces the equity supporting a position just as an adverse price move does. Depending on the venue and margin mode, it is taken from the shared account balance or from the position's own margin. Bybit, for example, deducts funding from available balance first and from an isolated position's initial margin only if that balance is insufficient, in which case the liquidation price moves closer to the mark price. Binance has likewise noted that funding is added to or subtracted from the wallet balance and can change the liquidation price.

Example 4: a funding stress test (hypothetical, simplified). Take an isolated long of 0.5 BTC entered at 64,000 USDT, and model liquidation as the mark price at which margin plus unrealized P&L falls to a flat 0.5% maintenance margin. Assume the price stays at 64,000 while funding runs at 0.10% per 8 hours for ten days (30 intervals), costing 30 × 32,000 × 0.001 = 960 USDT, all debited from the isolated margin because no other balance is available.

At 20x, funding alone consumed 60% of the margin and cut the distance to liquidation by two-thirds without any price move. At 5x, the same payments moved the estimate by the same amount, about 1,930 USDT, but left most of the cushion intact. By comparison, Example 2's 47.55 USDT of funding would move the 5x estimate by only about 96 USDT.

These figures show direction and scale only. Real liquidation depends on tiered maintenance margins, fees, the venue's mark-price methodology, margin mode and liquidation process, and a position can be closed before a simplified threshold is reached.

What funding rates can and cannot tell you

A persistently positive rate shows that perpetual buyers have been paying to hold long exposure, and a negative rate shows the reverse. That is information about positioning on one venue over one window, not a reliable signal of where price goes next. Rates can stay elevated for long periods, can differ across venues for the same asset, and can shift because of interval or cap changes that have nothing to do with sentiment.

Some traders try to collect funding by pairing a perpetual with an offsetting position. Those trades carry their own basis, execution, collateral and liquidation risks and are covered in a separate guide. On their own, funding receipts are compensation for holding a position that others are paying to avoid, not a yield.

Key takeaways

• Positive funding means longs pay shorts; negative funding means shorts pay longs. Payments pass between traders.
• On Binance, Bybit, OKX and Hyperliquid the rate is the premium plus a clamped interest term, which, with the default interest term, pins it at the equivalent of 0.01% per 8 hours whenever the average premium sits between −0.04% and +0.06%.
• Each payment equals the notional at the reference price at the funding moment times the rate. Entry price and leverage do not enter the calculation.
• Intervals, quoting conventions, reference prices and accrual methods differ by venue and change over time.
• Holding costs depend on the path of rates and prices; constant-rate estimates can understate them.
• Annualized funding is an indicative conversion, not a forecast or a realizable return, and it is far larger when measured against margin.
• Inverse contracts pay funding in the coin, so accumulated funding and collateral carry price exposure.
• Funding debits reduce equity and can move a leveraged position meaningfully closer to liquidation.