Perpetual Futures

A perpetual future is a futures contract with no expiry. It delivers leveraged, linear exposure to an underlying asset without a fixed settlement date, and it uses a periodic cashflow (funding) instead of a roll to stay anchored to its reference price.

Perps are the dominant derivative for altcoins and for everything outside the small set of assets with deep listed-options markets. On Hyperliquid, they account for the majority of open interest and daily volume across every asset class the venue supports. The contract is simple in spec and dense in consequences, so it pays to understand what you actually hold.

The Core Idea

A standard listed future has a last trading day. Before that day, a trader closes the position or rolls it into the next month's contract. The expiry is what forces the futures price to converge to spot: at settlement, the future and the underlying have to be the same number.

A perpetual future has no last trading day. So that convergence mechanism is gone. Something has to replace it, or the perp's price will drift away from its reference.

That something is funding: a periodic payment between longs and shorts, sized to the gap between the perp's traded price (the mark) and its reference price (the oracle). When the gap is positive, longs pay shorts. When the gap is negative, shorts pay longs. Either way, holding the wrong side of the gap costs money, which gives somebody a reason to take the other side.

Funding state:

Positive Funding: Perp trades above spot. Longs pay shorts. This mechanism pushes the perp price back toward spot, because the side that's "winning" the divergence has to keep paying for it.

ℹ️

The one-sentence definition

A perpetual future is a leveraged, linear exposure to a reference price, where funding replaces the roll as the mechanism that pulls the mark back toward the reference.

The Four Prices on a Perp

A perp has four prices you need to understand, and they are not all the same number.

Price	What it is	Where it comes from
Mark price	The last trade on the venue's order book	Live matching on the venue
Index / oracle price	The reference the perp is supposed to track	Published by the venue (weighted median of spot CEX prices, a curve blend, or a builder-provided feed)
Fair price	The mark smoothed by a short-window EMA, used for liquidation checks	Computed locally by the venue
Last funding rate	The rate paid at the most recent funding interval	Derived from the premium index over the interval

The premium is the difference between mark and oracle, usually quoted in basis points:

$$\text{premium} = \frac{\text{mark} - \text{oracle}}{\text{oracle}}$$

If premium is positive, the perp trades above its reference. If negative, below. Premium is the input that determines funding.

Position Mechanics

Long vs short

A perp has two sides. Every open long on the venue is offset by an equivalent open short. Total long OI equals total short OI at all times.

Side	Profits when	Pays funding when
Long	Mark goes up	Premium is positive (mark > oracle)
Short	Mark goes down	Premium is negative (mark < oracle)

Unlike options, payoff is linear in both directions. A 10% move up makes the long 10% (times leverage); a 10% move down costs the long the same amount. There is no built-in convexity.

Leverage and margin

Perps let traders take a position larger than their account balance by posting margin:

$$\text{position\_notional} = \text{margin} \times \text{leverage}$$

At 10x leverage, $1,000 of margin controls a$10,000 notional position. The same 1% adverse move that would cost an unlevered trader $100 now costs the levered trader$100 against $1,000 of margin, which is a 10% hit to their equity.

Liquidation

If an adverse move pushes the account into maintenance-margin territory, the venue forcibly closes the position. Hyperliquid expresses this by tracking a maintenance margin number defined as equity minus a per-position maintenance requirement, and triggers liquidation when that number goes negative:

$$\begin{aligned} \text{maintenance\_margin} &= \text{equity} - \sum_i |\text{notional}_i| \times \text{maintenance\_margin\_rate}_i \\ \text{liquidation triggers when } &\text{maintenance\_margin} < 0 \end{aligned}$$

The maintenance margin rate varies by asset but is typically 1–5% of notional. When the trigger fires, the venue tries to close the position on the book; if it can't be closed before the position goes underwater, the loss falls back to auto-deleveraging, which forcibly closes profitable counterparties at the previous mark.

⚠️

Liquidation is an order, not a price

Liquidation doesn't happen at an exact mathematical price. It happens when the venue's risk engine notices equity is below maintenance margin and submits a market order to flatten. In a fast market, that fill can be well below where the liquidation was "supposed" to trigger. Slippage is your loss, not the venue's.

Perp vs Spot vs Future vs Option

Four ways to get exposure to the same underlying, four different objects.

Perpetual Future

Linear leverage, no expiry

• No expiry — hold as long as margin is intact
• Leverage up to the venue max
• Funding paid every hour or every 8 hours
• Linear exposure — gains and losses are symmetric
• Liquidation and ADL risk

Best for directional traders who want leverage and do not want to think about expiry.

Spot

Linear, unlevered, owned outright

• You own the asset
• No leverage, no funding, no liquidation
• Storage, custody, and on-chain gas costs apply
• Linear exposure — same as perp but without the hourly bill
• Can be lent or deployed in DeFi for yield

Best for long-term holders who do not need leverage or a hedge.

Listed Future

Linear, dated, tracks a curve

• Fixed expiry date, hands you a dated contract
• No funding — carry is in the curve (contango or backwardation)
• Roll required if you want continuous exposure
• Cleared on a centralized exchange with margin
• The reference converges to spot at expiry

Best for institutional hedgers and anyone who wants predictable carry.

Option

Convex, dated, nonlinear payoff

• Pays a premium upfront for a conditional payout
• Convex — gains can exceed premium, losses are capped at premium (if long)
• Theta decays as expiry approaches
• Leverage comes from the premium-to-notional ratio, not margin
• See [Options vs Perpetuals](/docs/reference/options-vs-perps) for the full comparison

Best for traders with a view on how much the underlying will move, not just the direction.

What the Oracle Actually Tracks

The interesting question on any perp is: what does the oracle reference? The mechanic is always the same (funding pulls mark toward oracle), but the object the oracle points at changes everything downstream. A well-referenced BTC perp is boring. A perp referenced against a dated futures curve in the middle of a roll window is anything but.

Real examples, from the simplest to the most exotic:

Spot indices

The most common oracle on a crypto perp: a weighted median of spot prices from a handful of liquid centralized exchanges. The weighting is usually by liquidity, sometimes by recency.

Perp	Venue	Oracle
BTC	Hyperliquid, Binance, Bybit, OKX, dYdX	Weighted median of BTC-USD spot across the top 4–6 CEXes. Updated every 2–5 seconds.
ETH	Hyperliquid, Binance, Bybit, OKX	Same template as BTC but for ETH-USD.
SOL, HYPE, most altcoin perps	Hyperliquid	Same template, sometimes with fewer reference venues when the underlying is only listed on 2–3 spot markets.

Funding on these perps is usually small and reflects sentiment. Mark and oracle stay within a handful of basis points most of the time. The "negative funding = crowded short = squeeze" intuition was built entirely on perps like these.

Inverse / self-referencing (the BitMEX legacy)

BitMEX's original XBTUSD is an inverse perp: it pays out in BTC instead of USD, so the margin and PnL move with the underlying. The oracle is still a USD spot index, but the contract's own unit of account is the crypto itself. This is why "inverse perp funding" and "linear perp funding" look different in practice even though the formula is identical.

Perp	Venue	Oracle
XBTUSD (original 2016 perp)	BitMEX	.BXBT index (Bitstamp + Coinbase weighted)
ETHUSD inverse	Deribit, BitMEX	ETH-USD spot index, margined in ETH

Futures curves (the commodity/equity perp case)

A perp whose oracle references a dated futures contract. Funding absorbs whatever carry exists in the underlying curve. If the front-month is in contango, funding is positive. If in backwardation, funding is negative. The perp's reference itself can drift with no sentiment input from the perp market at all. Multiple HIP-3 builders deploy these with slightly different oracle specifications:

Perp	Deployer	Oracle	Note
xyz:CL (WTI crude)	trade.xyz	Front-month NYMEX WTI future (CLK6, etc.)	The canonical example. Funding hit -878% annualized during the April 2026 oracle roll.
xyz:BRENTOIL	trade.xyz	Front-month ICE Brent future	Thinner book than xyz:CL, funding behaves similarly.
km:USOIL	Kinetiq	Kinetiq's own spot-oil oracle, not a dated CME contract	Tracks a continuous WTI reference instead of a dated curve. Funding behavior looks more like a normal crypto perp than a commodity one.
flx:OIL	Felix	Felix's internal oracle (a blend of multiple spot feeds)	Smaller book, tends to pin tightly to oracle with thin depth. Behaves like a demo of the mechanic more than a real carry market.
xyz:GOLD	trade.xyz	COMEX front-month gold future	Small, usually well-behaved funding.
xyz:SILVER, xyz:COPPER, xyz:ALUMINIUM	trade.xyz	Front-month LME or COMEX contract	Same mechanic, different underlying.
xyz:CORN, xyz:WHEAT	trade.xyz	CBOT front-month	Seasonal curves make these interesting.

Note that xyz:CL, km:USOIL, and flx:OIL are three different oracle designs on the same underlying asset (WTI crude). The same spot oil price can produce wildly different funding regimes depending on which oracle the perp references. The trade.xyz version walks on a scheduled curve blend; the Kinetiq and Felix versions track a continuous spot reference with no roll mechanics at all. That divergence is the whole reason the oracle matters more than the mark on these instruments.

Curve blends (the rolling oracle)

A variant of the futures-curve case: instead of pointing at a single dated contract, the oracle tracks a published blend of two adjacent contracts. During the roll window, the weights shift from the expiring month to the next month in scheduled steps. The perp's reference itself walks on a published schedule.

This is the mechanic at the center of the Hyperliquid oil perps story. trade.xyz walks the xyz:CL oracle from 100% May into 100% June in five 20% steps, one per trading day, over about a week. When the underlying curve is backwardated, every step drags the reference lower even if crude itself does nothing. Shorts earn the reference drop, longs pay for it, and funding goes to extremes because the carry is predictable, not speculative.

Equity and index oracles

Twenty-four-hour perps against equity indices and single stocks. The oracle usually prioritizes live external pricing during market hours and falls back to an internal reference during off-hours, since the underlying itself stops trading at 4 p.m. ET.

Perp	Venue	Oracle
XYZ100 (S&P 500 tracker)	trade.xyz on Hyperliquid	During market hours: SPX / SPY. Off-hours: an internal discovery window bounded by the previous close.
xyz:AAPL, xyz:NVDA, xyz:TSLA, etc.	trade.xyz	Per-ticker spot during market hours, internal oracle off-hours.
xyz:EUR, xyz:JPY, xyz:DXY	trade.xyz	FX reference rates with 24/7 coverage.

Historical oddities worth knowing

Perp	Era	What the oracle did
BitMEX XBTUSD	2016–	First real perp, first funding rate, still trades. The template every other perp copied.
FTX MOVE contracts	2019–2022	Not technically a perpetual future, but a daily-settled volatility contract that behaved like one. Reference: absolute daily BTC move. Died with FTX.
Perpetual Protocol v1 vAMM	2020–2021	Pure virtual AMM. No reference oracle at all — the "price" was the AMM's internal state, and funding kept it tethered to an external index via a separate mechanism. Deprecated in favor of v2.
dYdX v3 cross-margined perps	2021–2023	Standard spot-index oracle, but with StarkEx L2 settlement and an off-chain order book.
Hyperliquid hyperp (hypothetical perps)	2024–	Experimental: perps with no external oracle at all. The funding rate is determined relative to a moving average of the perp's own mark. Useful for pre-launch token price discovery.

⚠️

Always check the oracle before you trade the perp

For BTC and ETH perps, the oracle is usually uninteresting. For everything else — commodities, equities, HIP-3 builder markets, any perp with "xyz:" or "km:" or "flx:" in front of the symbol — the oracle specification is the first thing to read before you click buy.

The Two Things That Replace the Roll

A perp needs two distinct mechanisms a listed future doesn't:

1. Funding — the periodic cashflow that pulls the mark toward the oracle when the two drift apart. Paid every hour on Hyperliquid, every 8 hours on most CEXes. This is the main tool.
2. Auto-deleveraging — the last line of defense when a liquidated position goes underwater and the insurance fund can't cover the loss. Profitable traders on the opposite side get forcibly closed to balance the book.

Together, these keep the perp solvent and anchored. Funding handles the ordinary mismatch; ADL handles the catastrophic one.

Key Things to Watch

If you are actually trading a perp, there are four numbers worth tracking:

1. Mark minus oracle — the premium. Directly sizes the funding payment you'll get or pay at the next interval.
2. Current funding rate — annualized, it's the cost of carry. Compare against your expected holding period and expected PnL.
3. Maintenance margin utilization — how close your equity is to the liquidation threshold. Above 70% is dangerous.
4. ADL rank — only matters if you're on the profitable side of a large trade. Higher profit and higher leverage put you closer to the front of the queue.

None of these are on the "buy" button. You have to go look them up before you open the position and keep watching them while you hold.

A Brief History (and why crypto, not Wall Street)

The perpetual future is younger than most people assume, and its real-world version is entirely a crypto story. It was proposed in academia more than a decade before any venue actually traded one, and every instance running today lives on a crypto venue. The reason has very little to do with technology and everything to do with why traditional finance never needed it.

Why it didn't come from Wall Street

Robert Shiller sketched the idea in his 1992 book Macro Markets: an instrument with no expiry, settled via periodic cashflows, that would let investors express exposure to hard-to-trade assets like real estate indices or GDP. He formalized the math more rigorously in a 1993 Journal of Finance paper. It was a serious academic proposal. Nobody listed it.

The reason is that traditional futures markets did not have the problem that perps are designed to solve. CME, ICE, Eurex, and the LSE already offer dated contracts across every liquid asset class, with regulated settlement, established market makers, physical delivery infrastructure, and a mature ecosystem of rolling strategies (calendar spreads, carry trades, basis arb). A professional futures desk does not mind the monthly roll. The roll is where they make money. Adding a perpetual variant would have meant extra regulatory approval for a contract type with less operational clarity and no obvious institutional demand, since the existing futures market already served every use case those clients cared about.

Traditional finance also has no need for 24/7 price discovery on a single contract. CME closes. The market knows how to handle that. Weekends and overnight gaps are a solved problem in workflows and risk systems. A perpetual oil future adds complexity without solving anything that wasn't already fine.

Why it came from crypto

Crypto in 2016 had none of that. Spot crypto markets traded 24/7 with no futures infrastructure, no clearinghouses, no physical delivery (obviously), and no venue willing to put a new contract type through years of clearance review. BitMEX was serving a retail audience that had never traded a dated contract and did not care what one was. The constraints that made perps unattractive in TradFi did not apply, and the product fit was better:

• No futures curve to live in. There was no deep BTC futures curve in 2016. A perp was the first leveraged derivative on BTC, not the simplified version of a more mature product.
• 24/7 markets, 24/7 contracts. Crypto never closes. A contract that also never closes matches the market's cadence. You do not have to schedule a roll around a weekend that does not exist.
• No delivery story. There is no crypto equivalent of "settling" a Bitcoin contract the way WTI settles into physical oil at Cushing. Cash settlement is the only option, and cash settlement plus a funding rate is equivalent to a perp.
• Retail-friendly. Dated contracts require the user to understand expiry, roll mechanics, and calendar spreads. A perp requires them to understand one number: the funding rate. The simpler product won with the retail audience, and from there grew to institutional.
• Creative freedom. The first-generation crypto venues could ship a new derivative, watch whether it worked, and iterate in weeks instead of years. That speed is the reason the funding-rate mechanism exists at all. Arthur Hayes and his co-founders tried several variations before settling on the one that actually kept the perp tethered to its reference.

Once BitMEX demonstrated that the funding-rate mechanism worked, the rest of crypto copied it. Binance added perps in 2019, Bybit, OKX, and FTX followed, and onchain versions (dYdX, Perpetual Protocol) appeared in 2020. By 2023, perps were the dominant leveraged derivative in the entire crypto market by volume, outstripping spot by a factor of 5–10x on most venues.

Timeline

Year	Event	Why it mattered
1992	Shiller proposes "perpetual futures" in Macro Markets	First serious academic argument for an instrument with no expiry, settled via cashflows rather than delivery. Never traded.
2016	BitMEX launches XBTUSD inverse swap	The first real perpetual future to trade, built by Arthur Hayes, Ben Delo, and Sam Reed. Inverse-margined in BTC. Introduced the funding rate mechanism the whole market now uses.
2019	Binance Futures launches USDT-margined perps	Made perps accessible to a much larger audience by margining in stablecoins instead of the underlying. Quickly became the dominant perp venue by volume.
2020–2021	OKX, Bybit, FTX, dYdX, Perpetual Protocol ship competing perps	Crypto converges on the perp as the default leveraged derivative. BitMEX's market share collapses. Onchain perps (dYdX v1, Perpetual Protocol) begin to find product-market fit.
2022–2024	Hyperliquid launches and scales	Order-book-based onchain perps with CEX-grade performance. First onchain venue to credibly compete with centralized perps on volume and latency.
2025	Hyperliquid ships HIP-3: builder-deployed perps	Third parties (trade.xyz, Kinetiq, Felix) can deploy their own perps with custom oracles, leverage, and fee structures. Unlocks commodities, equities, FX, and other non-crypto underlyings.
2026	Commodity and equity perps on HIP-3 become the largest books on the venue	For the first time, a DEX is the largest 24/7 price discovery venue for instruments that have their own deep traditional markets (oil, stocks, indices). See the April 2026 oil perps insight.

ℹ️

The loop closed

The perpetual future was invented because crypto had no futures infrastructure. Ten years later, the biggest books on the biggest crypto perp venue reference dated CME futures. A contract type built to avoid traditional finance is now the most active way to trade traditional finance assets 24/7. This is a real shape change, and it is the context for almost every interesting perp story being written right now.

See Also

• Perpetual Funding — the funding formula, premium index, and how the payment is actually calculated on Hyperliquid.
• Auto-Deleveraging (ADL) — how ADL queues work, who gets forcibly closed, and why it matters for basis traders.
• Options vs Perpetuals — the structural comparison: linear vs convex, funding vs theta, liquidation vs expiry.
• Hyperliquid Funding Docs — the official specification of the funding formula.
• HIP-3: Builder-Deployed Perps — how third-party builders deploy perps with custom oracles on Hyperliquid.