The Pre-IPO Perpetual Is a Window, Not a Forecast

The pre-IPO perpetual contract for Unitree on Trade.xyz last reported at $87.525. That is roughly 590 yuan per share. The official STAR Market IPO price is 150.8 yuan per share. The gap is not a spread. It is a 3.91x assumption. Multiply the pre-IPO price by the post-issuance share capital of 404 million shares, and the derivative market is assigning Unitree about $35.4 billion in market capitalization, roughly 238.7 billion yuan. The IPO itself will issue 40.4464 million shares, exactly 10% of post-issuance shares. One lot is 500 shares. The subscription payment is 75,400 yuan. If the pre-IPO contract is correct, those 500 shares are worth 295,000 yuan after listing. The implied profit: 219,600 yuan. The implied return: 291%. I audited the void and found a backdoor. The backdoor is not in Trade.xyz's contract. It is in the assumption that a pre-IPO perpetual price and a subscription lottery payout are the same number. They are not. One is a marginal synthetic price. The other is a conditional cash flow that depends on allocation probability, listing-day liquidity, and oracle mechanics. Unitree is a Chinese robotics company building quadruped and humanoid robots. It is not a crypto company. That is exactly why this trade is interesting. Unitree's IPO subscription opens tomorrow. The pricing venue that is emphasizing the opportunity is not the Shanghai exchange. It is an on-chain derivatives platform called Trade.xyz, listing a pre-IPO perpetual. A perpetual contract is a synthetic instrument with no expiry. In normal crypto markets, a perpetual tracks an existing spot index through funding rates. Here, there is no spot price until the shares list. So the contract trades against a reference that does not exist yet: the future listing price. That absence of a firm anchor is the defining condition of this market. The buyer and the seller are not exchanging shares. They are exchanging a promise to settle against something that may exist, in some form, after the IPO. This is the new shadow market for IPOs. In the old system, you got pre-IPO exposure only through a private allocation, a broker's balance sheet, or a high-priced secondary block. In the new system, a smart contract opens a synthetic order book before the official market opens. The quote is transparent, the funding rate is visible, and the order book is global. That is real innovation. But innovation is not the same as accuracy. A pre-IPO perpetual is a discovery mechanism, not a valuation mechanism. It discovers what the marginal buyer is willing to pay for exposure to a highly anticipated event. It does not discover the fundamental value of the enterprise. The two can be close. They can also be 3.91x apart. Before treating 291% as a fact, I had to check whether the contract is even pricing the same asset. The typical pre-IPO perpetual has three possible settlement designs. It can settle on the official IPO price, the first day's open, or an oracle chosen after listing. Each design creates a different basis. The numbers in the news are based on the last reported Trade.xyz price at $87.525. If the settlement reference is the IPO price, then the premium is a pure sentiment. If the settlement reference is the first trade or a volume-weighted average, then the premium embeds an estimate of how much the price will run in the early session. The quoted '291% potential profit' is simply the arithmetic difference between the pre-IPO perpetual and the subscription price. It is not an arbitrage. In a normal arbitrage, you hold the asset and the hedge until the spread closes. Here, a long perpetual is not a substitute for a share. It does not give you voting rights, dividends, or the right to sell into the IPO. It gives you a cash-settled bet. A subscriber who buys the perpetual is not locking in the spread; the subscriber is doubling the same directional exposure. If the listing opens below the subscription price, both the perpetual and the IPO allocation lose value. The math is symmetrical. The only situation where 291% is real is the one in which the listing actually prints above the perpetual price and you were allocated shares at the IPO price. Let me be explicit about the lottery. The subscription payment is 75,400 yuan per lot. The IPO plans to issue 40.4464 million shares, or about 80,892 lots. The demand is likely to be massive. In a STAR Market IPO, oversubscription is common. If the final oversubscription is 100x, your chance of winning a single lot is roughly one in 100. The expected profit from a single lot is 219,600 divided by 100, which is 2,196 yuan. That is 2.91% on the frozen subscription amount, before the refund delay. If the oversubscription is 1,000x, the expected profit is about 220 yuan per application, or 0.29%. The pre-IPO perpetual does not know that you won the lottery. It is pricing the share, not your probability. The 291% is a conditional payoff, not an expected return. The same math applies to the market cap. At the IPO price, the implied market value is about 60.9 billion yuan, or roughly $9 billion. The pre-IPO perpetual is marking the company at 238.7 billion yuan. That is $35.4 billion before a single share trades. The difference is not evidence of a better company. It is evidence of a more aggressive marginal buyer. The IPO will raise 6.1 billion yuan, roughly $900 million, based on the 40.4464 million shares offered at 150.8 yuan. So the issue will raise less than three percent of the market capitalization implied by the pre-IPO contract. The company itself gets the issue price. The entire 291% premium is a matter of primary-market allocation and secondary-market redistribution. It is not a transfer to the issuer. That is a structural clue: the premium is not money moving to Unitree. The premium is money moving between traders. The float is the next variable. A 40.4464 million share offering is 10% of the post-issuance company. A 10% public float is thin. With the issue price at 150.8 yuan, the free-float market cap is about 6.1 billion yuan. But the pre-IPO perpetual is marking the total market cap at 238.7 billion yuan. That means a $35 billion company has only 10% actual tradable supply. In a market with 10% float, large moves require only a fraction of the notional. The 3.91x premium may be less about underlying business value and more about float scarcity. I have seen this pattern before. In NFT markets, floor sweeps did not establish true valuation; they established a price for the lowest-priced asset in a thin book. Floor sweeps are just data points in motion. A last trade on a pre-IPO perpetual is the same kind of data point. It tells you where a small order transacted, not where broad supply and demand balance. The market cap extrapolation is dangerous. A price of $87.525 times 404 million shares equals $35.4 billion. That is a linear extrapolation from a derivative last price. No broker in the world will do a $35 billion valuation analysis based on that quote. The quote is the result of retail order flow, not a comprehensive shareholder offer curve. If the real supply of shares were 100% free-floating, the perpetual would need much smaller premium to clear the market. The 3.91x is a scarcity premium, a risk premium, and a lottery premium wrapped into one number. Each component has a different decay schedule. Scarcity decays as the free float expands in future unlock events. Lottery premium decays as the subscription deadline passes. Risk premium decays as the listing price becomes observable. The $87.525 number is not a stable target. It is the intersection of several decaying curves. Now the oracle issue. Smart contracts execute truth, not intent. That sentence is a slogan, but it is also the design principle. The problem is that a pre-IPO contract's oracle is not an objective natural law. It is a chosen reference. If the oracle is the first print on the STAR Market, then the contract will settle on a single transaction. That first print can be small, aggressive, or negotiated. If the oracle is a volume-weighted average of the first 30 minutes, then the contract is exposed to manipulation during the most fragile market-state transition in the listing process. A smart contract can enforce a rule perfectly and still execute a false signal, because the contract's truth is the oracle's truth. The bug is not in the code. The bug is in the choice of what the code treats as truth. I need to be honest about my own bias. In 2020, I spent two months reverse-engineering Curve's stableswap invariant and found a slippage exploit in its peg model. The loophole was not in a function's arithmetic. It was in the invariant's reaction to volatility. I learned that the most dangerous contract is the one that behaves exactly as specified. The same is true here. The pre-IPO perpetual on Trade.xyz is behaving as specified. The problem is the specification of the underlying reference. The contract will probably work exactly as intended. The intended reference may not be the same as the price that a reasonable equity analyst would assign. There is a lesson from my 2017 EOS arbitrage work: real inefficiencies have a structural cause. In late 2017, I found a latency gap in the EOS presale token distribution and wrote a C++ bot that predicted block production times with 98% accuracy. The edge existed because the system was deterministic and the settlement was in code. The pre-IPO perpetual is the opposite. Its settlement depends on a future market with human intermediaries, exchange rules and a regulator. You cannot code a backdoor around that. Cost of carry is another layer that the 291% headline ignores. When you buy a pre-IPO perpetual, you are long a synthetic position and paying funding to the short side. The funding rate exists because there is no borrowable share supply. If the market remains hot, funding will be paid by longs. That ongoing payment bleeds the 291% gap before listing. The gap between $87.525 and the IPO price is not your profit. It is a funding curve. It will be mined by arbitrageurs who do not care about Unitree. If the listing is delayed, the funding flow increases and the premium compresses. The finite window between now and the first print is the only period in which the perpetual can sustain an inefficient premium. After the first print, the reference becomes real and the basis evaporates. What does 3.91x look like historically? The STAR Market has seen enormous listing-day gains and dramatic losses. Single-day pops of 100% to 200% happen. A 3.91x gap implies the market expects a listing-day gain of roughly 291% from the issue price. That is not unprecedented in retail-driven markets, but it is not a normal equity event. For a robotics company with real revenue and real competition, a 3.91x premium is an aggressive assumption. It says the last marginal buyer in the pre-IPO market is expecting to sell to someone even more desperate on day one. That is a chain-letter property, not an investment property. The institutional angle matters here. Traditional institutions do not need a public chain to price Unitree. They have bankers and comparable-company analysis. Their participation in a STAR Market IPO is governed by rules, not by funding ratios. The price premium that retail traders are participating in is not a signal from institutional capital. It is a retail shadow market that is importing IPO demand into synthetic dollar-denominated tokens. The on-chain price discovery is real as a phenomenon, but its representativeness is questionable. The fact that the quote appears on a blockchain does not make it more truthful. It makes it more accessible. Accessibility and truth are different variables. Here is the contrarian read. The popular view is that a pre-IPO perpetual at 3.91x the issue price proves the listing is going to rip. The contrarian view is that the perpetual is already the sell-side. The product being sold is not shares. It is the dream of a 291% gain. The buyer of the perpetual pays the premium, pays funding, and assumes oracle risk. The seller, by contrast, is not short a real share. The seller is short a lottery ticket. If the listing disappoints, the seller collects the premium and the funding. If the listing surprises, the seller pays out a number that was already embedded in the quote. The asymmetry favors the seller. This is not a firm-specific criticism. It is the structural property of synthetic pre-IPO markets. The retail subscriber is the one with the asymmetric opportunity, not the perpetual holder. The subscriber can only win if allocated. The perpetual holder can only win if the reference goes above the entry price plus funding paid. That is a higher bar. The claim that the perpetual implies a 291% potential profit is inverted. The perpetual is the instrument that is already carrying the 291% expectation. It is not a forecast of what you will make. It is the cost of what you will have already paid if you enter at that level. Let me run a concrete scenario. Assume you are a subscriber with one lot. You pay 75,400 yuan. The reference listing price, according to the perpetual, is about 590 yuan. If you are allocated and sell at 590 yuan, you collect 295,000 yuan. Profit: 219,600 yuan. But you do not know your allocation probability. If the oversubscription is 100x, your expected profit is 2,196 yuan. If the oversubscription is 1,000x, your expected profit is 219.6 yuan. The 291% number is real only for the winners. The probability-weighted return is what matters for positional sizing. The same logic applies to any pre-IPO strategy: the edge is not the listing premium. The edge is the product of the listing premium and the probability of receiving the asset. The perpetual gives you the first number. It says nothing about the second. There is also the question of whether the perpetual can even be held through the listing. In many synthetic markets, trading is suspended or settled before the underlying event. If the contract is structured to settle at a fixed reference, the long holder does not get to choose the exit price. The reference is selected by a rule, not by human discretion. If the reference is the opening auction price, then the contract unlocks at the exact moment when price discovery is most chaotic. The holder is forced to accept the settlement as truth. Smart contracts execute truth, not intent. But the truth in question is the oracle's truth. Before the first trade, the oracle is nothing. The 'truth' is whatever the last perpetual holder is willing to accept. That is intent, not truth. The last issue is liquidity depth. The last reported price of $87.525 is a single quote, not a robust index. In a market with little depth, one buyer can sweep the ask and push the last price to an extreme. In my NFT trading work, I learned to distinguish between a floor price and a transaction. A floor price is a quote. A transaction is an event. The Trade.xyz last price is an event, not a distribution. Building a 291% expected-value case on a single event is statistically indefensible. You need to see the order book, the funding rate, the open interest, and the spread before you can treat the price as meaningful. A perpetual with a wide bid-ask spread and tiny open interest can report a number that no one can actually trade at size. The 2022 Terra collapse taught me a different lesson. The issue was not simply that TerraUSD was an algorithmic stablecoin. The issue was that the system needed price to go up in order to remain solvent. A pre-IPO perpetual with a 291% premium has the same property. Its value depends on the IPO continuing to exceed the previous quote. If that feedback loop breaks, funding will turn and the contract can gap faster than the underlying. The underlying share always has an official price. The perpetual has only a memory of its last trade. That memory can be erased by one aggressive sell order. The market is telling you the window is open, not that the window will stay open. If you get allocation, sell into the first liquid bid. Use the perpetual as a risk gauge, not as a target. And if the perpetual's funding rate is positive and rising, respect it. The math can be gamed, but the void does not lie. I audited it. The backdoor is the settlement reference, the funding curve, and the lottery probability. None of those are hidden. They are simply not included in the headline. The question is not whether the IPO is a 291% arbitrage. The question is whether you can collect it before the reference price becomes a fact. The first print will be a truth. Will it be your truth?