Inside the Perpetual: The Mechanics of Funding Rates

Introduction

Futures are legally binding agreements between two parties: one agrees to buy, and the other agrees to sell a specified commodity or financial instrument at a predetermined price on a set future date. This structure allows speculation on a crypto asset's price without direct ownership.

Classical futures contracts have an expiry date. At settlement, the futures price is forced to converge to the underlying spot price. Any deviation is resolved mechanically by the settlement process.

To see why this is so, suppose that the futures price is above the spot price during the delivery period. Traders then have a clear arbitrage opportunity: 1) sell a futures contract, 2) buy the asset in the spot market, and 3) make delivery. These steps lock in a profit equal to the amount by which the futures price exceeds the spot price.

Perpetual (perp) futures are different. They never expire. There is no settlement date where the market is forced to close the gap between perp and spot. While this feature is convenient for traders uninterested in rolling contracts every month, it introduces an awkward problem.

Just because a contract is called "BTC perpetual" (BTC perp) doesn't mean its price would naturally stick to BTC spot. Bitcoin Cash is a good example: it shares the name and history of Bitcoin, but in practice, it trades as a separate asset with its own price dynamics. Without a mechanism that forces convergence, a BTC perp could evolve the same way, just another BTC-branded instrument whose price is driven by its own supply and demand, not by the underlying spot market.

This is where funding comes in. Funding is what stops that and keeps perp and spot moving together. Funding is a continuous incentive system that nudges the perp price back toward spot by making the crowded side pay the other side.

The difference between the futures price and spot price is called Basis ($\text{basis} = \text{futures price} - \text{spot price}$). This difference may lead to a premium or positive basis when the futures price is higher than the spot, or to a discount when the futures price is lower than the spot.

Funding and Funding Rates

Funding is a periodic cash transfer exchanged between traders holding long and short positions in a perpetual contract.

Funding is not a fee for the exchange. The platform only facilitates the payment, which is exchanged between traders based on their positions. A positive funding rate means longs pay shorts; a negative one means shorts pay longs. This mechanism is what incentivizes prices to realign.

How Exchanges Determine Funding Rates

Different exchanges calculate and settle funding at different time intervals. For example, Binance computes funding every 8 hours, while Hyperliquid does it every hour.

In general, centralized exchanges determine the funding rate from four main ingredients: the premium, a base interest rate, an average premium index, and a clamp value that bounds the adjustment within a certain range.

A typical formula for the funding rate looks something like this (different exchanges have different variants):

$\text{Funding Rate} = \text{Average Premium Index} + \text{clamp}(\text{Interest} - \text{Premium}, -d, +d)$

The purpose of the clamp is to prevent extremely volatile or punitive funding rates. It caps the adjustment term $(\text{Interest} - \text{Premium})$ within a band (for example, from -0.05% to 0.05%), so that funding remains stable even if prices are jumpy.

The Premium

The premium is a normalized measure of how far the perp is trading from the underlying spot index. Roughly:

$\text{Premium} = \dfrac{\text{Mark Price} - \text{Index Price}}{\text{Index Price}}$

From the above expression, if the perp trades above spot, the premium is positive; if it trades below, the premium is negative.

Index Price

The index price is typically a composite spot price for the underlying, built from spot prices on several large exchanges. It is the best estimate of the "real" underlying spot price, frequently called an oracle. This avoids funding being driven by a single venue's local quirks or a temporary spike.

Mark Price

The Mark Price is a theoretical price representing the “fair value” of the futures contract. It is used for unrealized PnL calculations, funding settlements, and liquidation triggers. Because these are key drivers of market operations, the Mark Price must neither be overly sensitive nor too sluggish in reflecting market changes.

The Mark Price typically depends on the index price, the basis between the perp and the spot, and a volume-weighted average of recent order-book mid-prices.

The Interest Rate

The Interest Rate is a small base rate that reflects things like USD vs coin interest or generic cost-of-carry assumptions. Many venues fix this around 0.03% per day (scaled to the funding interval), with exceptions for specific pairs.

The Average Premium Index

The Average Premium Index is the time-weighted average of the premium over the funding interval. It is calculated by comparing the perpetual price to the spot price over a specific time period. For example, an exchange might compute an 8-hour average premium index using a formula that gives more weight to recent prices.

A Quick Numerical Example

Suppose BTCUSDT is trading at a mark price of 83,950 USDT (the spot price at the time I am writing this paragraph) and the current funding rate for the next interval is +0.01% (0.0001). If you are long 1 BTC perpetual, your position value is:

$\text{Position Value} = 1 \times 83{,}950 = 83{,}950\ \text{USDT}$

The funding fee you pay for this interval is:

$\text{Funding Fee} = 83{,}950 \times 0.0001 = 8.395\ \text{USDT}$

If the exchange charges funding every 8 hours and the rate stayed at +0.01% for the whole day (3 intervals), you would pay about 25.2 USDT in funding over that day. Annualized, that corresponds to roughly an 11% cost for holding the leveraged long position, even if the BTC price itself went nowhere.

How Funding Pulls the Perp Back to Spot

Now that we know how funding is calculated, it is easier to see how it helps keep the perp price close to spot. The key idea is simple: when the perp trades at a premium to spot, funding is usually positive; when it trades at a discount, funding is usually negative.

Positive funding creates an incentive to short the perp and buy spot. Those trades push the perp price down and the spot price up, compressing the premium and, through the funding formula, lowering the funding rate itself. The process continues until some kind of equilibrium is reached.

In a simple, frictionless world, the equilibrium is where the funding rate is just high enough to compensate you for the cost of financing the spot inventory you hold against your short perp. If funding is above that level, more arbitrage flows in; if it is below, capital leaves the trade.

Example of the Equilibrium: a Toy Model

Suppose that:

• BTC trades at a spot price of 84,000 USDT,
• the perpetual future also trades at 84,000 USDT (for simplicity, we assume perp and spot are equal at this instant),
• you can borrow USDT at 0.005% per 8 hours (about 5.5% per year),
• the funding rate is currently +0.03% per 8 hours, so longs pay shorts.

In these circumstances a trader can:

• borrow 84,000 USDT at 0.005% per 8 hours,
• buy 1 BTC in the spot market at 84,000 USDT,
• short 1 BTC perp at 84,000 USDT.

The trader's PnL over the next funding interval is approximately:

$\text{PnL} \approx 84{,}000 \times 0.0003 - 84{,}000 \times 0.00005 = 25.2 - 4.2 = 21\ \text{USDT}$

This 21 USDT is near-riskless carry for 8 hours (ignoring fees and slippage), so many traders will try to put on the same trade. Their collective selling of the perp and buying of spot compresses the premium, and the funding formula then produces a lower funding rate.

As funding falls toward 0.005% per 8 hours, the net carry of the trade falls toward zero. In the idealized limit where funding exactly equals the financing cost, the expected PnL of adding more size to this basis trade is roughly zero, so new arbitrage capital stops coming in. That's the equilibrium in the Toy Model: perp and spot sit close together with a funding rate just high enough to pay for the cost of carrying spot inventory.

In the real world, you also have to subtract trading fees, the cost of financing the cash or collateral posted as perp margin, balance-sheet constraints, and an additional risk premium for holding inventory. As a result, the true equilibrium funding rate typically sits a bit above the pure financing cost: just high enough to compensate arbitrageurs for all of these frictions, not just the spot borrow.

Calculating the Premium and Funding: Hyperliquid Example

Hyperliquid's way of calculating the premium and the funding rate is different now than it was when the exchange launched in 2023. I am writing this in January 2026, so I will first describe the current formulas and then work through an example.

Throughout this section, when I say "current", I mean "as of late November 2025". Hyperliquid may change these details again in the future, so always check the official docs if you are reading this much later.

The Current Formulas (November 2025)

Hyperliquid's official formulas for the premium and funding can be found here: https://hyperliquid.gitbook.io/hyperliquid-docs/trading/funding.

Premium

The premium is given by:

$\text{premium} = \dfrac{\text{impact\_price\_difference}}{\text{oracle\_px}}$,

where:

$\begin{aligned} \text{impact\_price\_difference} &= \max(\text{impact\_bid\_px} - \text{oracle\_px}, 0) \\ &\quad - \max(\text{oracle\_px} - \text{impact\_ask\_px}, 0) \end{aligned}$

In summary, the impact price difference will be the impact bid price minus the oracle price when both the impact bid and impact ask prices are above the oracle price. Conversely, the impact price difference will be the impact ask price minus the oracle price when both the impact bid and impact ask prices are below the oracle price.

Impact bid and ask prices are “depth-aware prices”: the average price to trade a fixed notional on each side of the book. Hyperliquid uses them instead of just the top-of-book to compute the premium for funding, so funding reflects the true cost of pushing size through the order book, not just a single quote that can be spoofed or illiquid.

More precisely, on Hyperliquid:

• Impact Bid Price = the average fill price you’d get if you sell a fixed notional (the “impact notional”) into the bids.
• Impact Ask Price = the average fill price you’d get if you buy that same fixed notional into the asks.

Currently, Hyperliquid’s funding impact notional is 20,000 USDC for both BTC and ETH. So:

• BTC Impact bid price = average execution price to sell 20k USDC worth of BTC across the bid side.
• BTC Impact ask price = average execution price to buy 20k USDC worth of BTC across the ask side.

Funding rate

The formula for the funding is:

$\text{Funding Rate} = \text{Average Premium Index} + \text{clamp}(\text{Interest} - \text{Premium},-0.0005,+0.0005)$

According to Hyperliquid's documentation, the Average Premium Index is calculated by sampling the premium every 5 seconds and averaging it over the hour. However, for both of the tables I use as examples, the premium column can be exactly recovered by applying the premium formula, and the funding rate can be exactly recovered by using this premium directly in the funding formula, instead of the Average Premium Index.

The interest rate component is fixed at 0.0001 every 8 hours, or 0.0001/8=0.0000125 per hour, or 0.0003 per day (1.0003³⁶⁵ − 1 ≈ 11.6% annually) and the clamping is 0.0005.

More, simply:

$\text{clamp} = \min\bigl(\max(0.0001 - \text{premium}, -0.0005), 0.0005\bigr)$

The funding is then:

$\text{funding}_{8h} = \text{premium} + \text{clamp}$

For compatibility with other exchanges, these values correspond to 8-hour periods, but Hyperliquid funding is done every hour. So, to get the actual hourly funding rate, it is necessary to divide the funding rate by 8.

$\text{funding}_{1h} = \dfrac{\text{funding}_{8h}}{8}$

Current Data Example

Hyperliquid's public data (which they call asset_ctx – meaning asset context, I presume) includes all the variables necessary to understand how they calculate the premium and the funding rate.

The following table shows a subset of the most recently available context data sampled daily at 0:00 UTC, from 2025-10-22 to 2023-10-31.

It is important to note that, as the current data show, it is entirely possible to have a positive funding with a negative premium.

For example, on 2025-10-31 0:00 UTC:

time:          2025-10-31 00:00
oracle_px:     108314.0
impact_bid_px: 108360.0
impact_ask_px: 108361.0
premium:       0.000425

$\text{impact\_price\_difference}=\max(108360-108314,0)-\max(108314-108361,0)=46$

$\text{Premium} = 46/108314 \approx 0.000425$

$\text{clamp} = \min\bigl(\max(0.0001 - 0.000425, -0.0005), 0.0005\bigr) = -0.000325$

$\text{funding}_{1h} = (0.000425-0.000325)/8 = 0.0000125$

Historical Formulas

Historically, at the launch of Hyperliquid, both premium and funding had slightly different formulas. Also, at least at the very beginning, the funding rate was presented as an 8-hour rate.

Premium

At the launch of Hyperliquid, the premium was given by the following formula:

$\text{Premium} \approx \dfrac{\text{impact\_mid} - \text{oracle\_px}}{\text{oracle\_px}}$,

where

$\text{impact\_mid} = \dfrac{\text{impact\_bid\_px} + \text{impact\_ask\_px}}{2}$

Funding Rate

The funding rate formulas were:

$\text{clamp} = \min\bigl(\max(0.0001 - \text{premium}, -0.0003), 0.0003\bigr)$,

and:

$\text{funding}_{8h} = \text{premium} + \text{clamp}$

Historical Data Example

Let's use BTC's oldest data as an example. The following table shows a subset of the context data sampled daily at 0:00 UTC, from 2023-05-21 to 2023-05-30.

2023-05-21 0:00 UTC, example:

time:          2023-05-21 00:00
oracle_px:     27104.0
impact_bid_px: 27079.4
impact_ask_px: 27098.5
premium:      -0.000555

$\text{impact\_mid} = \dfrac{27079.4+27098.5}{2} = 27088.95$

$\text{Premium} = \dfrac{27088.95-27104.0}{27104.0} \approx -0.000555$

$\text{clamp} = \min\bigl(\max(0.0001 - (-0.000555), -0.0003), 0.0003\bigr) = 0.0003$

$\text{funding}_{8h} = (-0.000555+0.0003) = -0.000255$

Conclusion

For classical futures, the combination of cost-of-carry arbitrage and delivery at expiry keeps futures and spot prices tethered from listing to settlement: if futures wander too far from spot plus carry, arbitrageurs have a clear trade to push them back in line.

Perpetual futures remove the hassle of rolling contracts: they have no expiry and no delivery. This means they can't rely on that hard convergence. Instead, they use the funding mechanism to play a similar role, softly nudging the perp price back toward the spot index over time.

Under the surface, the mechanics are simple: a premium that measures how far the perp trades from spot, a base interest rate, an average premium index, and a clamp that prevents extreme values. Wrapped around this is an incentive structure: when the perp is rich, longs pay shorts; when it is cheap, shorts pay longs. Arbitrageurs respond to those incentives, and their flows tend to pull the perp back toward the spot index.

In that sense, funding is the ongoing price we pay for the convenience of a future that never expires.

This blog is not an invitation to trade. Also, it does not intend to provide trading advice. Trading involves risks, and readers are solely responsible for their trading decisions.

Trading futures is inherently a high-risk activity. You can lose more than your initial margin. This material is for educational purposes only.