Bybit added Unitree and Moonshot AI to its pre-IPO perpetuals lineup this week, pushing its TradFi perpetuals range past 200 products spanning equities, ETFs, commodities, indices and private companies. Over on Hyperliquid, traders are pricing the robot maker at close to $38 billion against a $9 billion IPO price, according to Allium data reported by CoinDesk on 15 August 2026. That is roughly four times, on a company whose shares have never traded once.
There is nothing scandalous about betting on what a private company will be worth. What is worth stopping on is the vocabulary. A perpetual future on a stock that has no public price is not price discovery. It is a wager settled against a number that somebody, somewhere, gets to define. And almost nobody buying one can tell you who that somebody is.
TL;DR
- Bybit listed pre-IPO perpetual futures on Unitree and Moonshot AI, part of a TradFi perps lineup now above 200 products.
- Hyperliquid traders value Unitree at about $38 billion versus a $9 billion IPO price, a gap that cannot be arbitraged because the underlying shares cannot be borrowed or shorted.
- A perpetual on a private company has no continuous public spot price, so the mark price that triggers your liquidation is a methodology chosen by the venue, not an observed market.
- Satoshie runs raffles and coinflip on Base with odds fixed in the contract and winner selection from Chainlink VRF, where the randomness arrives with a cryptographic proof verified on-chain before the contract accepts it.
- The lesson is not that pre-IPO perps are fraud. It is that crypto already built verifiable settlement, then went back to selling numbers nobody can check.
What a pre-IPO perpetual actually is
A perpetual future has no expiry. It stays tethered to its underlying through a funding rate: when the contract trades above the index, longs pay shorts, and the economics drag it back. That mechanism only works if the index means something. On a listed stock, it does. The index is built from continuous public quotes across venues that anyone can watch, and if the contract drifts, an arbitrageur trades the real shares against it until the gap closes.
On a company that has never listed, there is no continuous public quote. So the index has to be constructed. Depending on the venue, it might come from the venue’s own order book, from occasional secondary market transfers, from the last primary funding round, or from some blend of those weighted by rules in a terms document. When the index leans heavily on the venue’s own book, the tether becomes circular: the contract is anchored to a price the contract itself produced.
That is a defensible design. It is also a completely different risk from the one the interface implies, because the interface shows a chart, a leverage slider and a liquidation price, exactly as it would for Apple.
The 4x gap is not a forecast
Treating a $38 billion mark against a $9 billion IPO price as the crowd’s genuine estimate of Unitree’s value gives the number more credit than it has earned. Thin books, no borrow, no way to sell the actual asset short, and a supply of enthusiastic leveraged longs produce exactly this shape. In a normal market a fourfold premium gets attacked by anyone with access to the underlying. Here nobody has access to the underlying. The only thing that can close the gap is the listing itself, which is precisely why the analysts flagged that leveraged positions get dangerous the moment real trading starts.
The market is not wrong about Unitree. The market has not been allowed to have an opinion yet.
The question nobody in the thread is asking
If you hold one of these contracts, your position does not die because you were wrong. It dies because the mark price touched a level. So the honest questions are: what feeds that mark, how are the inputs weighted, and who can change the weighting. What happens if the IPO is delayed by a quarter, or pulled entirely, or comes as a direct listing rather than a book build. What is the settlement rule if the listing prices below the last private round. What happens if the venue amends the methodology while your position is open.
Every one of those answers usually lives in a terms page that can be edited after you have taken the trade. That is the actual product: a contract whose defining parameter is revisable by the counterparty.
What a coinflip does differently
Here is the part that reads like marketing until you check it, so go and check it. Satoshie runs raffles and coinflip on Base. The odds are written into the smart contract and published before anyone plays, and they cannot be adjusted for a particular player, a particular round or a bad week. Winner selection comes from Chainlink VRF: the contract requests randomness, and the random value arrives with a cryptographic proof that is verified on-chain before the contract will accept it. If the proof fails, the result is rejected. Not by us. By the chain.
Settlement is immutable and public. Every draw, every input, every payout stays visible long after anyone cares about the outcome. There is no methodology page, because there is no methodology to revise. The mechanism is the disclosure.
A coinflip is a smaller thing than a pre-IPO derivative, obviously. That is the point. The smaller product carries the higher standard of proof, and the larger one asks for trust.
The bar is lower than the industry pretends
None of this requires a venue to stop listing pre-IPO perps. It requires four things: publish the index constituents and their weights, publish the settlement rules for every listing outcome before the product goes live, commit to not changing either while positions are open, and put both somewhere a user can verify rather than somewhere a lawyer can revise. Any venue could do that this quarter. The reason most will not is that the ambiguity is worth money.
Meanwhile the same industry that lists 200 leveraged products on assets with no observable price still describes provably fair on-chain gaming as gambling. Fine. One of those two things publishes its odds and proves its randomness. The other shows you a number and asks you to size your position around it.
The tell
When a product will not show you its mechanism, the number on the screen is the marketing. Unitree may well be worth $38 billion. It may be worth a third of that. Nobody trading the perp is going to find out from the chart, because the chart is not measuring the company. It is measuring the enthusiasm of everyone else holding the same contract, priced by a venue that also runs the book.
On-chain gaming settled this argument by refusing to have it. Show the odds, prove the randomness, make the record permanent, and there is nothing left to take on faith. It works for a coinflip. It would work for a derivative too, if anyone selling one wanted it to.
Try a provably fair game at satosh.ie and verify the result yourself on-chain.
📷 Photo by Gabriele Malaspina on Unsplash
