Pre-Launch Perps: Trading a Price That Doesn't Exist Yet

What a pre-launch perp is missing

A pre-launch perp — Hyperliquid calls its version a hyperp, Binance and other exchanges list them as pre-market contracts — is a perpetual future on a token that does not trade yet. The token has been announced. It may have an airdrop list, a mainnet date, an IEO or IDO reference price. What it does not have is a spot market.

That absence removes something more fundamental than liquidity. Every other market answers three questions the same way: what is the fair price, how is it discovered, and what is the index. On an ordinary perp the answers come from outside the contract — the index is a basket of spot prices from several exchanges, discovery happens on those spot books, and fair value is whatever they agree on. Take the spot market away and all three answers have to come from inside the contract itself.

Hyperliquid's documentation states the design constraint plainly: the contract "does not require an underlying asset or index that exists at all points of the hyperp's lifetime, only that the underlying asset or index eventually exists for settlement or conversion." That is the whole product in one sentence. You are trading a claim whose reference price is scheduled to begin existing at some point, and until then the market is the only evidence of what the market thinks.

If perpetual mechanics are new to you, perpetual futures explained covers the ordinary version first. Everything below is what changes when the spot leg is removed.

Where the price comes from when there is no spot

Both venues arrive at the same unavoidable answer — the contract references itself — and then diverge completely on how to make that safe.

Hyperliquid replaces the oracle with a lagged average of its own mark. The external spot/index oracle price is replaced with an eight-hour exponentially weighted moving average of the last day's minutely mark prices. The published formula runs over 1,439 one-minute samples with an exp(−i/480) decay, and — this is the part that matters for a brand-new market — "when there are fewer than 480 mark price samples, the initial mark price is used as the padding value." The whole expression is then capped at four times that initial mark price.

Binance does not construct an index at all. During pre-market trading the mark price is the "average of last 10 seconds trade prices, calculated every second," falling back to the last 20 transaction prices when fewer than 21 trades occurred in the interval. On top of that sits a hard leash: a ±1% cap on how far the price may move per second.

Why the first price the market prints matters more than it should

On an ordinary market the opening print is just the first observation. Here it is written into the contract.

Until Hyperliquid's market has accumulated 480 minutely samples — eight hours of trading — its oracle is padded with the initial mark price, so for the whole first session the reference price is anchored to whatever the market opened at. That initial mark also sets a ceiling: the oracle can never exceed four times it. On Binance the mechanism is different and the effect is similar. With the mark defined as a trailing ten-second trade average and a ±1%-per-second leash, the level the first aggressive orders establish is the level everything afterwards has to walk away from, one percent at a time.

This is why early pre-launch prices are structurally different from post-listing prices rather than merely noisier. A launch auction, an IEO price or a TGE valuation does not feed either venue's formula as data — but whoever trades against that number in the first minutes sets an anchor that the contract's own machinery then defends for hours.

Why arbitrage cannot anchor the price, and funding has to

On a normal perp, the tether to spot is not really funding. It is the trade funding pays for: when the perp trades above spot, someone buys spot, shorts the perp, collects the funding and waits. That trade is riskless enough to be crowded, which is why the basis stays small. The mechanics are in funding rates explained.

Remove the spot market and that trade cannot be constructed. There is nothing to buy, nothing to borrow, nothing to deliver, and no date on which the two prices must meet. Whatever premium the pre-launch perp carries, no one can lock it in — they can only take the other side and hope. So the risk premium in a pre-launch perp is not an arbitrage spread that got briefly wide. It is a genuine payment from whoever wants exposure to whoever is willing to be short something they cannot hedge.

Each venue then chooses how hard to lean on funding to keep that premium finite.

Hyperliquid prices the imbalance. Its funding rate is P + clamp(interest rate − P, ±0.05%), paid hourly, capped at 4% per hour, where the premium P is measured from the impact bid and ask against the oracle price. For a hyperp, that oracle is the contract's own eight-hour average — so funding is charged on the gap between where the contract trades now and where it has been trading. It is a momentum toll, not a basis. The documentation warns about exactly this: "if there is heavy price momentum in one direction, funding will heavily incentivize positions in the opposite direction for the next eight hours."

Left unmodified, that toll would be ruinous. So hyperps scale it: "funding rate premium samples are computed as 1% of the usual clamped interest rate and premium formula." Substituting the two published rules into each other shows what that scaling is worth.

The right-hand column reaches Hyperliquid's 4%-per-hour cap at any premium above roughly 32% — below that it is simply the premium spread over eight hours, which is why the 10% row is not at the cap. So a contract pinned at its own 3x mark cap would charge a long about 96% per day under the ordinary formula, and about 6% per day with the pre-launch scaling applied. That is the trade the design makes: funding stays survivable, and in exchange it stops being a strong enough force to drag a dislocated price back.

Binance rations the imbalance instead. Its pre-market funding rate is not premium-driven at all — it is fixed at 0.005% per four-hour interval, roughly 1.1% a year, for the entire pre-market period. Funding is doing no anchoring work whatsoever. The constraints are the ±1%-per-second speed limit and the position tiers: 5x leverage only up to 5,000 USDT of notional, and above 80,000 USDT you are limited to 1x. You cannot get a big enough position to push the market far, and you cannot push it fast.

What the caps actually protect against

Hyperliquid runs three ceilings at once: the mark is capped at 3x the eight-hour mark EMA, capped again at 1.5x the median external pre-launch perp price where such a market exists, and the oracle itself is capped at 4x the one-month average mark.

Read them together and their shape becomes clear. Every one of them is a multiple of a lagging average of the contract's own price. They bound how fast a dislocation can arrive, not how large it can eventually become — because as the average catches up, the ceiling rises with it. A violent hourly spike is contained. A determined push sustained over days is not, because the thing being pushed is also the thing that sets the limit.

The one genuinely external term is the third mark-price component. Hyperliquid's documentation is unusually candid about the compromise involved: "despite the often significantly different contract specifications between hyperps and other venues' pre-launch perp markets, they are nonetheless included as mark price inputs to provide greater mark price stability during volatility." Another venue's pre-launch contract is not an index — it has its own leverage limits, its own funding, its own mark construction, and its own crowd. It is imported anyway, on the reasoning that a noisy outside opinion beats a purely self-referential one.

The XPL case: a price the token never reached

The clearest demonstration of what all of this means for a position ran through one token over five weeks in late 2025.

The pre-launch market. Late on Tuesday 26 August 2025, Hyperliquid's XPL/USD hyperp spiked to nearly $1.80, corrected, then spiked again toward $1.00 the following day. Binance's pre-market contract on the same token, over the same period, reached only $0.55. Two simultaneous prices for the same unlaunched token, 3.3x apart, with no arbitrage available to close the gap — neither contract was deliverable, and neither had a spot leg to hedge into.

What it cost. More than $17 million of positions were liquidated, predominantly short positions, after four addresses swept the order book for a roughly 2.5x move and reportedly took more than $46 million out of it. Ordinary liquidations were not enough to absorb the move and the venue shifted to auto-deleveraging within minutes. Hyperliquid subsequently said it would cap hyperp mark prices at a multiple of the eight-hour EMA and pull in external pre-launch perp data — both of which appear in today's documentation as the 3x cap and the CEX median component — while noting that the change "would not have changed any of today's liquidations or ADLs."

What happened next. XPL began trading with the Plasma mainnet launch on 25 September 2025, and made its highest price ever inside the first week. CoinGecko records the all-time high as $1.68, dated 27 September 2025. On Binance specifically, the highest daily close was $1.63 on 27 September 2025 and the highest intraday print $1.693 the following day. Every one of those numbers is below the $1.80 the pre-launch contract had printed a month earlier. Figure 1 plots the whole path from the first day of spot trading onward, and the walkthrough beneath it steps through what to read off each marker.

Reading Figure 1

Walk the chart from its left edge. The series starts on 25 September 2025, the day spot trading opened — the first marker, and the first candle the venue has. The second marker, two days later, is the highest daily close the token has ever made: $1.63. Everything to the right of it is a decline. As of 12 August 2026 XPL trades at about $0.075, some 95.5% below the all-time high, per CoinGecko.

Mind the difference between the two peak numbers. The chart plots closes, so its high point is $1.63; CoinGecko's $1.68 all-time high is an intraday figure aggregated across venues, and Binance's own highest intraday print was $1.693. That spread of six cents does not change any argument here, and naming it is the point — three defensible answers to "what was the highest price," all of them well under $1.80.

Now place the pre-launch prints against it. The $1.80 Hyperliquid print of 26 August 2025 is above every price on this chart, and above both intraday highs, and it happened a month before the chart begins. The $0.55 Binance print from the same hours sits well inside it. The pre-launch market did not merely misjudge the level; one venue's version of it quoted a price the asset has never since reached.

The consequence for a position is the uncomfortable part. A trader short at $1.80 in August 2025 had the correct view by any standard you like: XPL never traded there, and it trades 96% lower today. They were liquidated anyway, and the ADL queue closed what the liquidation engine could not. Being right about the eventual price is not what protects a margin position — the mark is, and in a pre-launch market the mark is whatever the order book can be pushed to.

What to do with it: size a pre-launch position against the mark's worst plausible path, never against your view of fair value, because those are different numbers with different consequences and only the first one can liquidate you. Concretely: if a 2.5x spike against you would take your margin, then the position is too large regardless of how wrong that spike would be — the XPL book delivered exactly that in a single session. Cross-margined exposure to a market whose mark references itself is the version of this error that also reaches your other positions.

Funding, open interest and the listing move

The usual pre-listing readouts are funding, open interest and pre-market volume. All three exist for these contracts. What they mean is not what the same fields mean on an ordinary perp.

Open interest is not comparable across venues here. On a Binance pre-market contract, open interest is bounded by construction: 5x maximum leverage and a 1x limit above 80,000 USDT of notional put a ceiling on how much position the market can hold. A hyperp on the same token has different limits, a different mark, and a different funding regime. Comparing the two numbers, or adding them, compares two differently-shaped containers. Open interest explained covers what the field measures when the contracts are alike; that precondition fails here.

Funding is a positioning readout, not a basis readout. This is the point from the section above, restated where it bites: elevated hyperp funding tells you the contract is above its own recent average, which is information about crowding and momentum rather than about richness to fair value. It cannot tell you the market is expensive relative to the listing price, because no price for the listing exists yet.

The catalyst calendar is the part that transfers. Listing dates, airdrop distributions and the first unlock cliffs are real, dated supply events, and they behave the way they do elsewhere — vesting schedules and token unlocks is the mechanism, and memecoin odds explained is the sobering base rate for new launches. What a pre-launch perp adds is that you can now be positioned for those events with leverage, well before the float that will absorb them exists.

How the position actually ends

There is no expiry date and nothing is delivered. That surprises people who arrive expecting a convergence trade, so it is worth stating as a sequence.

The practical implication is that the classic listing-gap trade — short the rich pre-launch contract, wait for the real price — has no enforcement mechanism behind it. You are not waiting for a settlement that must happen on a known date; you are paying funding, posting margin against a self-referential mark, and hoping the substitution arrives before the mark takes your position. XPL is the case study in what happens when it does not.

When nobody has a price at all

Pre-launch tokens are the easy case: a spot market is genuinely coming. The same machinery now carries markets where the reference asset may have no public price for years, or ever — pre-IPO equity being the obvious example.

Hyperliquid's builder-deployed perpetuals (HIP-3) make the arrangement explicit. The deployer of a market is responsible for "market definition, including the oracle definition and contract specifications," and for "market operation, including setting oracle prices, leverage limits, and settling the market if needed." The deployer stakes 500,000 HYPE, maintained for at least 183 days after deployment, and each such market has independent margining and order books.

So on a builder-deployed market the reference price for something with no public price is a value the deployer posts. That is not a criticism of the design — someone has to name a number, and the staking requirement and the deployer's identity are the accountability on offer. It is a reason to read the market's own specification rather than assuming the norms of the main venue apply to it.

Risks and limits

Conclusion

A pre-launch perp is an ordinary perpetual future with its anchor removed. Because no spot market exists, the venue has to manufacture a reference price out of the contract itself — an eight-hour moving average of its own mark on Hyperliquid, a ten-second trade average on Binance — and every consequence follows from that circularity. Arbitrage cannot operate, so the premium longs pay is a real risk premium rather than a spread someone will close. Funding stops measuring richness to spot and starts measuring momentum against the contract's own history, which is why Hyperliquid scales it to 1% of the normal formula and Binance replaces it with a constant. The caps that bound dislocation are themselves multiples of a lagging average of the price being bounded, so they slow a spike without limiting where a sustained push can end.

XPL is what that looks like from inside a position: $1.80 on one venue and $0.55 on another in the same hours, $17 million of mostly-short liquidations, and a highest-ever traded price of about $1.69 — below the pre-launch print, on a token now 95.5% below even that. The traders who were right were liquidated by a mark, not by the market's verdict.

None of which makes the instrument unusable. It makes the discipline specific: read the venue's mark and oracle definitions before the ticker, size against the mark's worst path rather than your view of fair value, keep the position isolated, and treat the conversion date as something the venue will decide rather than something you can plan around.

Frequently asked questions