Pre-Launch Perps Are Float Markets, Not Token Markets

The trade everyone thinks they're seeing

Pick any pre-launch perp on Aevo or Hyperliquid in the weeks before a token generation event. Multiply the implied price by the project's total supply. Compare the number to the FDV at which the last private round cleared. Almost always, the pre-launch number is materially below the round number. Sometimes by 2x. Sometimes by 5x.

The standard read on this is that the venture rounds overpaid. The crowd is wiser than the cap table. Smart money on the perp is calling the bluff on dumb money in private markets.

It is a clean story. It is also a category mistake. The pre-launch perp is not pricing the same thing the round priced. Once that's clear, the apparent gap stops being a verdict on anyone and starts being a mechanical artefact.

The frenzy, in numbers

It's worth grounding this in what's actually trading right now, because the magnitudes do most of the argument's work.

Jupiter prestocks. Jupiter Perps on Solana now lists synthetic perpetuals on a roster of private companies, with Anthropic, OpenAI, SpaceX, and Stripe as the marquee names. The Anthropic prestock has traded most of this month at implied market caps between $400B and $600B against the $1.4T secondary mark. The OpenAI prestock sits at a similar fraction of the consensus secondary range. Daily open interest on the top three names is in the low nine figures. Funding rates have been positive almost continuously since launch, with the Anthropic contract averaging in the 200-400% APR range and short-dated spikes well above that whenever a new secondary headline hits. The crypto-native commentary on these prices is essentially uniform: the public market is pricing these companies at half what private buyers think they're worth. That conclusion is what this essay is arguing against.

Hyperliquid pre-launch perps. Hyperliquid pre-lists liquid perpetuals on tokens weeks before TGE, with settlement to a TWAP of the post-TGE reference price. Funding on pre-launch markets is uncapped relative to spot perps, and the hottest contracts in the current cycle have spent extended periods at hourly funding rates above 0.1%. Annualised, that is over 900% APR. Peak hourly prints on the most one-sided books have touched 0.5% an hour, which annualises beyond 4000%. The HYPE pre-launch on its own platform in late 2024 famously implied a token value materially below where the token actually traded on day one. The same pattern has repeated on most pre-launches of the last twelve months: implied FDV at 40-60% of last-round FDV, settlement at or above the last-round implied price, and funding payers underwriting the discount for as long as it persists.

Aevo and Whales Market. Aevo has been the venue of record for pre-launch perps since 2024, with Eigen, Wormhole, and a long tail of L2 and infra tokens having traded materially below their eventual TGE TWAPs. Whales Market, which runs OTC physical settlement against actual token delivery, has consistently priced the same names higher than the Aevo and Hyperliquid perps on the same underlying. Different product, different settlement, different number. The convergence of all three only happens at TGE, and even then only partially.

These are not anecdotes. They are the regularity. Across most launches of the last two years, the pre-launch perp curve has been a structurally biased forecaster of the post-TGE clearing price, and the structural bias is exactly what you'd predict from the mechanics described below.

What pre-launch perps actually settle to

On Aevo and Hyperliquid, a pre-launch perp settles into a regular perp once the token lists. The conversion price is a time-weighted average of post-listing spot, typically over the first several hours to first day. The pre-launch perp is therefore a forward on the first-day TWAP of the public spot market.

That phrase is doing a lot of work. The first-day TWAP is not the project's value. It is the clearing price of whatever supply happens to be unlocked on day one, against whatever demand happens to show up. For most launches, both of those quantities are dominated by mechanical factors that have nothing to do with the business behind the token.

Whales Market is the cleaner analogue. There, pre-launch is OTC. Counterparties agree on a token price, post collateral, and at TGE the seller physically delivers tokens. There's no TWAP, no perp basis, no settlement noise. It is the closest thing to a true forward on the project. And, importantly, prices on Whales Market have historically been higher than the implied price from Aevo and Hyperliquid pre-launch perps on the same token, sometimes substantially so. Different instrument, different price.

The float wedge

A typical token at TGE has 10 to 25% of supply in public float. The rest is in one of three buckets: investor and team allocations behind a cliff and then linear vesting, treasury and ecosystem reserves under multi-year locks, and airdrop allocations with their own vest schedule.

Of the float that is unlocked on day one, a meaningful slice is held by airdrop recipients with near-zero cost basis. The empirical regularity is that 30 to 60% of any airdrop tranche is sold within the first 48 hours of trading. Add allocated market-maker inventory under loan agreements, and the day-one supply that has to clear is several percent of total supply, hitting the book against whatever organic demand shows up.

The first-day TWAP is the price at which that supply clears. It is, by construction, a sale price for sellers who have to sell. It is not a valuation for the long-only buyer of the underlying business. Mechanically, this price should be below any sensible estimator of fundamental fair value, because it embeds a forced-seller discount that fundamental value does not.

When the pre-launch perp prices in that discount in advance, it is not telling you anything about the project. It is telling you what the market expects the forced-seller clearing price to be.

There is no spot, so there is no basis

The reason normal perps track spot at all is funding. If the perp drifts above spot, funding is positive, longs pay shorts, an arbitrageur shorts the perp and buys spot to collect the carry, and the basis closes. The mechanism is rough but it works because the spot leg actually exists.

A pre-launch perp has no spot. The token doesn't trade anywhere yet. If the pre-launch perp trades above where consensus thinks first-day TWAP will print, funding goes positive and there's no way to take the other side except naked. An arbitrageur can short the perp, but they have no hedge until TGE. Whatever they get is convergence risk, not a basis trade.

The implication is that pre-launch perp prices can persistently diverge from any unbiased estimator of where they will settle, in either direction, and stay there for weeks. There is no mechanical force closing the gap. The closest thing to one is liquidation cascades on overcrowded positioning, which is a sentiment mechanism, not a valuation one.

The funding rates in the current cycle make the missing-arb point as cleanly as you could want. When the Anthropic prestock on Jupiter sits at 200-400% APR funding for weeks, what that number actually says is that longs are willing to pay shorts the equivalent of two to four times the notional per year for the privilege of being long, and the shorts taking the other side have no spot to hedge into. They are short a naked synthetic on a private company. They are not arbitrageurs. They are speculators on the convergence of an oracle reference to an indicative private mark, with no time horizon and no obvious catalyst. In a normal perp market, that funding rate would close in days. In a prestock market, it doesn't close at all, because the trade that closes it cannot be put on. The same logic applies on Hyperliquid pre-launch perps, where four-digit annualised funding can sit on the tape for the entire duration of the pre-launch window. Funding is not telling you what longs and shorts actually believe. It is telling you what longs will pay in the absence of an arbitrage trade to put the structural shorts out of business.

Asymmetric flow on the curve

Consider the structure of who has reason to be long or short the pre-launch perp before TGE.

Long: anyone who thinks the launch will be well bid. Pre-TGE hype, exchange listings, retail rotation. There is no equivalent of a strategic long, because strategic holders are getting their tokens for free or close to it via the cap table. They don't need the perp.

Short: anyone who thinks the first-day TWAP will undershoot the perp. That includes a specific group, namely participants with informed views about first-day sell pressure: airdrop hunters running their numbers on the eligible cohort, market makers with insight into their own allocation and inventory plans, and anyone modelling the unlock curve closely. It also, in principle, includes anyone with non-public information about the launch playbook.

The structure is asymmetric. The long side is dominated by retail with a narrative. The short side has at least some participants with structural information about day-one mechanics. That asymmetry alone is enough to push the pre-launch perp below the consensus first-day TWAP. It is not a market failure, it is just what happens when one side of the book is mostly informed and the other side mostly isn't.

Settlement methodology is the product

Two pre-launch perps on the same underlying token are not the same product if they settle differently. The most consequential parameters:

TWAP window length. A perp that settles to the first hour of trading is a sharper bet on initial sell pressure than one that settles to a 24 or 48 hour TWAP. The longer the window, the more it averages out the airdrop dump and the closer it gets to a slightly fairer estimator of early-market consensus.

Reference venue. If settlement TWAP is computed off the largest centralised exchange, the first hours are dominated by airdrop recipients and short-term traders. If it's computed off an on-chain reference with thinner books, you are settling to a more easily moved number. The choice of reference venue is, in effect, a choice about which microstructure regime the perp will be anchored to.

Physical versus cash settlement. Whales Market trades require the seller to deliver tokens. That makes the implied curve a real forward on delivery risk plus token value, which is structurally different from a cash- settled TWAP. Whales Market prices reflect counterparty risk and the value of actually receiving tokens. The two cannot be expected to converge.

Forward-to-event versus open-ended perpetual. Aevo and Hyperliquid pre-launch perps are forwards on a specific event: the TGE TWAP. They have a built-in convergence date, even if it's uncertain. Prestocks on Jupiter have no such anchor. They are open-ended perpetuals referenced to an oracle that tracks indicative private-market prices, with no IPO date in sight for most of the underlyings. The funding mechanism is doing all the work of "pricing," because there is no future event that will force convergence. That makes prestocks the purest expression of the missing-arb problem. The Anthropic prestock could trade at $400B implied or $1.4T implied for a year, and there is no mechanical reason for the gap to close until either Anthropic lists, the oracle methodology forces a mark, or the market gets bored.

Treating all pre-launch perps as the same product because they share a name is the same mistake as treating CME bitcoin futures and Deribit perpetuals as the same product because they're both bitcoin derivatives. They aren't. A prestock on Jupiter and a pre-launch perp on Aevo are even further apart than that.

What the discount actually decomposes into

The gap between the implied FDV from a pre-launch perp and the last-round FDV is a sum of effects, each of which has nothing to do with whether the project is worth what it raised at.

Float adjustment. The perp prices the clearing price of a small unlock, not the project. Mechanically multiplying that spot by total supply double-counts the structural sell pressure built into the public float.

No-arbitrage premium. Without spot, the perp can drift. Sustained demand imbalance produces persistent dislocation. There is no mechanism to close it.

Asymmetric information. Informed flow from airdrop modellers and market-maker insiders sits disproportionately on the short side. The unbiased consensus would already be below fundamental value. With this asymmetry, the mid drifts further below.

Settlement noise. A TWAP over the first hours of trading is a small sample dominated by mechanical sell pressure. The perp is forward-pricing that sample, not the project.

Liquidity premium. Market makers quoting a pre-launch with no spot to hedge against need to price their inventory risk into the spread. That widens both sides of the book and skews the mid downward when retail flow is one-directional long.

Sum these and a 60% or 70% discount to last-round FDV is not anomalous. It is what the mechanism produces under most launch playbooks.

What the price actually tells you

The pre-launch perp is a forecast of the first-day TWAP. It is a useful forecast for one set of questions and useless for another.

It is useful for: anyone trying to size their TGE-day participation. If the perp is trading at $0.40 and you think first-day TWAP will print at $0.55, there's a trade. Anyone modelling the airdrop cohort's behaviour, the MM loan schedule, the exchange listing slate, has a real edge on a real question.

It is useless for: anyone trying to back out fundamental value. The pre-launch perp is structurally biased low relative to fundamental value, by the mechanisms above, and the bias does not unwind at TGE. It unwinds slowly over months, as unlocks happen and the float ratio normalises.

The mistake is reading the perp price as a valuation when it is a settlement forecast. The same chart, read as the first one, tells a story about VCs being wrong. Read as the second, it tells a story about airdrop dump expectations and MM inventory. Only one of those stories is what the instrument was designed to price.

The trades that actually work

Three patterns dominate among desks running real pre-launch books.

The first is the carry trade. Quote both sides of the pre-launch perp, collect spread and funding. The risk is convergence: if the perp gaps on settlement, the inventory turns into a loss before there's spot to hedge against. So the size has to stay small and the spread has to be wide enough to absorb that. The trade scales with how many pre-launch markets a desk can simultaneously run, not with how aggressively any one of them is quoted.

The second is the settlement-mechanics trade. In the last week before listing, the perp converges to first-hour TWAP. That last week is dominated by information about the eligible airdrop cohort, the MM allocation, the listing schedule. Desks with channels into those flows can take directional positions against the curve. This is where actual edge lives.

The third is the float-decomposition trade. Look at the tokenomics. Identify launches with unusually high first-day float, large airdrop allocations, or thin MM commitments. Those are mechanically biased to undershoot at TGE. Short the pre-launch perp, hold to settlement, take the difference. The trade has tail risk if the launch surprises to the upside, but on average the decomposition is correct.

None of these involve a view on whether the project is over or undervalued. They all involve a view on first-day clearing mechanics.

The grey-market analogue, and what Jupiter has actually built

For most of the last decade, the obvious comparison for crypto pre-launch perps was grey-market trading on Asian IPOs. Those grey markets have all the same features: no spot, settlement risk, asymmetric information from underwriters and allocated institutions, and a persistent discount to the eventual listing price. The grey market is a forecast of the open, not a valuation. People who trade it know this.

That comparison is now incomplete, because Jupiter prestocks are a product Western traditional finance has not built and crypto has. The Anthropic perp on Jupiter is doing, at small scale, what a pre-IPO derivative market on Anthropic shares would do if such a thing existed in equities. It has no spot. It has no IPO date. It references an oracle of indicative private-market prices. It clears billions in cumulative volume. Daily open interest sits in the nine figures. It is, functionally, the continuous derivative market on Anthropic shares that regulated equity venues have not been allowed or incentivised to construct.

And the price it prints is half the secondary mark, with 200-400% funding.

The first wave of commentary on this is, predictably, "the on-chain market is calling the bluff on Anthropic's $1.4T tender." The same commentary writes itself for OpenAI, SpaceX, Stripe. It is wrong for the same five reasons it is wrong about crypto pre-launch perps. There is no spot to arbitrage. Informed shorts have no offsetting long flow. Settlement methodology dominates the curve (in the prestock case, the curve is whatever the oracle and funding mechanism jointly produce, which is dominated by the structural short of the synthetic). The float that would clear at any real IPO is a small fraction of total shares, and the prestock is implicitly pricing that small float, not the company. And liquidity providers quoting a synthetic on a private equity with no hedge need a wide skew to absorb the inventory risk, which biases the mid down whenever flow is one-directional long.

The Anthropic prestock at $400B implied is not telling you the company is worth $400B. It is telling you what the funding market clears at when one side of the book is full of retail longs and the other side is mostly speculators shorting a synthetic that cannot be hedged. The number is real. It is also not what the headline reading takes it for.

The reason crypto has the product and equity markets do not isn't sophistication. It's that token markets have no legal distinction between primary issuance and secondary trading, and prestocks have so far slipped through the same gap by operating on Solana with oracle-referenced settlement and offshore venue structures. In US equities, every step of that stack is regulated separately. The mechanism is the same. The plumbing isn't, yet.

Crypto pre-launch perps and prestocks are, in one specific way, more honest than equity markets would be if they had the same product. The settlement methodology is published. The oracle inputs are auditable. The tokenomics or the private-mark inputs are in the open. There is no pretence of being a price-discovery mechanism for fundamental value, even when commentary on the same screen pretends otherwise.

Why this matters

Pre-launch perp volumes have grown into the billions of dollars per launch cycle on the larger venues. They are no longer a curiosity. They are a real product that real desks make markets in and that real capital trades through. The category mistake matters because pricing in one category and reading the result as another is how you end up with the wrong trade.

The recurring claim that pre-launch perps are evidence of private-market mispricing is the same kind of error as reading the front-month future on a commodity as a forecast of fundamental supply and demand. It isn't. It's a forecast of the settlement of a specific contract, with structural features that bias it relative to fundamentals. Treating one as the other is the trade that loses money.

What the pre-launch perp price is good for is what it was designed for: forecasting the first hours of post-TGE trading. That is a useful thing to be able to do. The desks doing it well are not trading project value. They are trading float, sell pressure, and the mechanics of listing day. The number on the screen is the price of that, and only that.

The Anthropic frame matters because the same misreading is already happening, now, in real time, on a venue that millions of people transact on. Jupiter prestocks have collapsed the gap between "the crypto pre-launch problem" and "the equity pre-IPO problem" into a single product running on a single chain. Forge and EquityZen offer fragmentary pre-IPO exposure on the regulated side, and the next generation of products on either side of the regulatory line will look much closer to a Jupiter prestock than to a tender. When they arrive at scale, the first wave of commentary will read the implied FDV as a valuation, the discount to the most recent secondary mark as a verdict, and the funding rate as information. None of those readings will be more correct in equities than they have been in tokens, or than they have been on prestocks for the last several months. The mechanism doesn't care which asset class it's running on. The people pricing it correctly will be the ones who watched Eigen and HYPE and Wormhole settle higher than their pre-launch perps implied, who paid attention to funding rates of 200% APR on an Anthropic synthetic with no hedge available, and who understood early that the number on the screen was a price of float and forced sellers, not a price of the underlying business.