12 min read

What Trade.xyz's Cerebras market tells us about the Pre-IPO summer

Source: http://x.com/i/article/2055310588813914112

Thank you to

*@hydromancerxyz*

,

*@HyperTracker*

(by

*@Coinmarketman*

),

*@castle_labs*

and

*@stacy_muur*

for their contributions to the research.

Introduction @cerebras (

$CBRS

) announced its IPO issue price at $185 a share on May 13th, against books reportedly 20x oversubscribed. Thirty million shares cleared at an implied market capitalisation above $48 billion, the largest US technology offering of the year. Only, a fraction of this allocation reached retail accounts.

The reason is structural. Underwriters distribute the bulk of any hot IPO to existing institutional clients: hedge funds, sovereign wealth funds, pension funds, and asset managers that route flow through the underwriting banks all year. Retail brokers receive a sliver, and the users of those brokers compete for that sliver through lotteries and minimum-balance gates. In a deal as hot as Cerebras, most retail accounts either received nothing or received insignificant allocations. The bulk of retail demand was routed through secondary markets where they bought the asset at a price well above where institutions were filled.

As a way around this limitation, sophisticated investors have historically turned to SPVs, or Special Purpose Vehicles. These are business entities formed for the purpose of pooling money from various investors to meet the minimum buy-in demands. While platforms like Hiive and Forge have made access to SPVs easier, they are generally not retail friendly. Further, with the news of Anthropic deeming them disallowed, the SPV workaround may be reaching the end of its lifespan.

https://x.com/lex_node/status/2053845294731641076

Prime brokers at the major banks write bespoke forward contracts on pre-IPO equity for hedge fund and family-office clients, but those require a prime-brokerage relationship, large credit lines, and seven-figure minimums.

From 01:30 UTC on May 1, when the contract was first quoted, through the evening of May 13, when Cerebras priced and the perp converged toward its post-conversion mark, the @tradexyz

CBRS perpetual was the only continuous retail-accessible CBRS price-discovery venue in the world.

It quoted through every meaningful event in the bookbuilding window: the S-1 amendment, the range hike, the upsize, the pricing guidance. By the time Cerebras priced, the market's twelve-day mid had absorbed every available datapoint and was quoting a price the underwriter's range could be measured against.

Image

Caption - Figure 1. Trade.xyz CBRS perpetual price trajectory from launch (May 1, $190) through pricing eve (May 13, $307), with vertical markers on the four events that defined the bookbuilding window: the S-1 release on May 4, the range hike on May 10, the upsize confirmation on May 11, and the pricing wire on the evening of May 13. The chart traces the contract's full pre-listing lifecycle in one frame and reads as the running argument between the underwriter's range and the market's continuously quoted price.

$CBRS is the first of many. The IPO pipeline through the next twelve to twenty-four months is dense with US tech and AI names: Stripe, Klarna, Databricks, Discord, OpenAI, xAI, Anthropic, and SpaceX are among the issuers expected to come to market. Each of these names will clear an institutional allocation that retail cannot reach, list at a Day-1 mark above the issue price, and leave most retail accounts buying into the secondary tape at a premium they did not get to set.

Pre-IPO perpetuals close this access gap. Retail-accessible, continuously priced, and deep enough to absorb the size that matters. They are, by some distance, the cleanest mechanism retail has to participate in this IPO summer on equitable terms.

https://x.com/FabianoSolana/status/2054972707025150246

What is a Pre-IPO Perpetual?

A Pre-IPO Perpetual (IPOP) is a cash-settled perp market that references the anticipated post-listing price of a pre-IPO company. Holders take exposure without holding a share, an allocation right, or any claim against the issuer. The instrument converts the discrete signals of the IPO process (filings, range updates, secondary prints, pricing wires) into a single continuous price.

A general prep market marks against an external oracle: a spot reference, futures basis, or index. A pre-IPO contract has no such oracle because the stock does not trade yet.

Trade.xyz solves this with an internal pricing mechanism similar to the one its other perps run over weekends, when reference markets are closed. The oracle is a thirty-minute exponentially weighted moving average that advances via the market's own previous prices.

For pre-IPO contracts,

trade.xyz applies a 0.005× funding multiplier on @HyperliquidX 's funding rate formula, 1% of the 0.5× multiplier on a standard HIP-3 equity perp. Funding exists to pull a perp back toward its fair value anchor, but a pre-IPO contract has no fair value to pull it toward. Suppressing the funding rate lets the market sit at large premiums or discounts for days without bleeding to carry, which is the only way the contract can express a forward view that diverges from any near-term reference price.

Discovery Bounds cap where the market can trade, to prevent a single order from running the book during low-liquidity periods while leaving headroom for the price to move on news events.

Once the underlying lists and regular-way trading begins on Nasdaq, the IPOP converts to a standard externally-priced equity perpetual with the conventional 0.5× funding multiplier. From that point, the perp marks against listed CBRS the same way xyz:TSLA marks against listed

$TSLA.

How

Trade.xyz

Tracked

$CBRS

's Listing Price

Trade.xyz deployed the CBRS perpetual at 01:30 UTC on May 1, three weeks before the listing date. The initial orderbook showed deployer-placed resting bids stepped between $50 and $125 with no asks. The first trade printed at $190 against the first deployer asks at 01:33 UTC. By the close of trading on May 3, three full days before the S-1 was amended, the contract had traded as high as $320.

The S-1 amendment landed on May 4 at 10:14 UTC with an initial range of $115 to $125, well below where the perp was already trading. The contract sold off to roughly $200 over the next twenty-four hours as informed leverage unwound, a move that ran through a two-hour cluster of more than nine hundred long-side liquidations during the 16:00 to 17:00 UTC window. By the end of the week it had ground back to $280 to $300.

The

Reuters scoop on the upsized range hit the wire on the evening of May 10 at 21:39 UTC: $150 to $160 against an enlarged 30M-share offering. The perp's reaction was effectively flat. It was already trading at $307, almost twice the new range high. The May 11 session confirmed the upsize and added a further $1.68M of fresh short interest from a cohort of fourteen named wallets. Price closed near $290.

On the evening of May 12

Bloomberg reported pricing guidance tracking above the $150 to $160 range. The perp drifted higher over the following twenty-four hours and closed at $307 on May 13 at 20:01 UTC. Pricing landed at $185 a share roughly an hour later. The contract moved from $310 to $284 in the eighty-nine minutes between the pricing wire and 21:29 UTC, an 8.4% adjustment on roughly 8,600 contracts of post-news volume.

https://x.com/castle_labs/status/2054974539835261082

Three venues offered continuous quotes on Cerebras through the bookbuilding window: the

trade.xyz

HIP-3 perpetual on Hyperliquid, the Day-1 closing market-cap binaries on Polymarket, and the SPV trades on Hiive. Each of these traded at different price levels.

Polymarket: the implied close price

@Polymarket was the only other retail accessible venue to take directional bets on the

$CBRS price apart from

@tradexyz, as it listed a "$CBRS closing market cap" market.

We decomposed the implied probability from this Polymarket into an estimate of expected close prices which requires assuming or forecasting Volatility of

$CBRS for its first trading day.

This was done by looking at prior oversubscribed AI/tech IPOs and assuming CBRS follows a similar volatility regime on its first trading day.

If you’re interested in reading the full methodology, check out our blog

here.

Hiive: the price after a lockup

Hiive price sit at the opposite end of the bookbuilding window. Through the last week of April, secondary trades on Hiive cleared at $102 to $107. The reason the price sat well below the perp's quote is structural rather than informational. Buyers on Hiive take delivery of pre-IPO shares, but are subject to long lockup terms. The holder cannot exit when the underlying first lists, instead having to . They have to wait six months, and sometimes longer, before the position is liquid.

Hiive was the only market of the three giving a multi-month inventory risk to the buyer, and the spread between the Hiive price and the perp mark is, almost entirely, reflective of that lockup.

Comparing the major venues

Image

Figure 2. Cross-venue price evolution for the trade.xyz CBRS perp, the Polymarket-implied Day-1 close (under 220% IV), and Hiive secondary prints across the bookbuilding window. Three series rescaled to share the same axis so the spread between them reads at a glance. The chart visualises the gap structure across instruments: trade.xyz quoting $190 to $320, Polymarket-implied tracking the $200 to $250 band, Hiive prints clustered at $100 to $110.

Image

Figure 3. The chart shows the $ spread of each venue to the VWAP price of

$CBRS on day 1 of its listing. Trade.xyz consistently priced in the closest with Polymarket-implied price and Hiive following after.

Image

Figure 4. Cross-venue convergence to the Day-1 close of Trade.xyz fills, indexed against the realised post-listing fills from Nasdaq.

Polymarket anchors on the Day-1 close, the

trade.xyz

IPOP anchors on the post-conversion mark, and Hiive prices a lockup-discounted secondary share. The cleanest single comparison across all three is Figure 3: each venue plotted as deviation from the realised Day-1 close, the variable each one can actually be measured against.

Trade.xyz was consistently closer to the realised post-listing price than either alternative, and it was the only one built to absorb the volumes that the market experienced during the IPO day.

Square profile picture

Arrakis @ArrakisFinance ·

May 14

Replying to

@ArrakisFinance

@tradexyz

and 3 others

The

@tradexyz market has corrected to a $340-350 level following major media outlets call for $CBRS to open at $350

Image

1

2

[

749

](/ArrakisFinance/status/2054949724017582493/analytics)

The

trade.xyz contract also reacted to news in real time. Bloomberg's report on the evening of May 12 that the range looked likely to clear above $160 was absorbed within minutes: orderbook quotes re-anchored higher, taker volume turned in the same direction, and the contract closed May 13th near $307 against an underwriter range still tracking $150 to $160. The same pattern repeated through the pricing wire itself, with the perp converging toward its post-conversion mark before the listing bell rang.

Image

Figure 5. Rolling realised volatility on the trade.xyz CBRS contract through the bookbuilding window, with vertical markers on each major news event. The chart shows the comparative quiescence around the May 10 range-hike wire (the move was already priced) against the visible volatility spikes on the May 4 S-1 release, the May 11 upsize confirmation, and the May 13 pricing wire (the largest single-event spike of the window).

Liquidity

Trade.xyz was the deepest CBRS book of the three venues by an order of magnitude, which is why it held its quote through the pricing-day volume surge.

For any trader with meaningful size, Polymarket was untradable. A few thousand dollar trade on the market would move the Polymarket book through more than 2% of impact to fill. The same trade on

trade.xyz filled at leas than 5 basis points consistently.

Image

Figure 6. Trade size required to incur 2% slippage on the trade.xyz CBRS perp versus the Polymarket CBRS binary, plotted as a daily series across the bookbuilding window. The trade.xyz line sits roughly two orders of magnitude above Polymarket through every regime, and the spread widens into pricing day as quoting intensity on the perp scales with news flow.

This is why $70.7M of total notional cleared across 208,854 fills on the perp over twelve days, and why the same window on Polymarket cleared a fraction of that. Depth on the perp held through every news event, and the contract absorbed institutional-scale volume without dragging its quote off the running fair value.

Image

Figure 7. Trades per minute on the trade.xyz CBRS perp across the bookbuilding window, with a 60-minute rolling average overlay. The chart shows steady-state activity around 5 to 15 trades per minute through May 1 to May 9, a step up to 30 to 60 trades per minute through the May 10 range-hike and May 11 upsize confirmation, and a peak of 80 to 120 trades per minute in the hour around pricing on May 13.

Image

Figure 8. Unique trader and holder counts on the CBRS perpetual, daily. Two series plotted: cumulative unique addresses that traded the contract (running total) and daily active holders (snapshotted at end-of-day). The cumulative line passes through 2,148 distinct addresses by pricing day. The daily holder count climbs steadily through the bookbuilding window and peaks the day before pricing. Trade.xyz was the only venue that had continuous quoting, retail-accessible execution, and a book deep enough for six-figure trades to fill without experiencing unbearable sippage.

The Winners and Losers

Market-structure analysis explains how the venue priced. It does not say who came out ahead. The participants who captured the largest leverage-adjusted returns over the bookbuilding window are the cleanest read on whether the access thesis held in practice.

We took the top ten wallets by percentage return on margin on xyz:CBRS. Every address was passed through Arkham Intelligence, with a recursive backtrace on the first inbound funder for any wallet that carried no direct identity tags.

Image

Eight of the ten wallets carry a retail-shaped Web3 identity profile. The remaining two have no on-chain identity beyond a centralised exchange deposit. None of these ten resolved to an institutional desk, a known market maker, or a fund.

This is the access asymmetry from the introduction showing up in realised PnL. While Cerebras allocation cleared overwhelmingly to clients of the underwriting banks at $185 a share, the continuous price-discovery venue that ran in parallel to that allocation placed its biggest leverage-adjusted gains in wallets that match the demographic the underwriting process screened out. The traditional shape on a hot IPO is that institutions receive the allocation and retail buys the secondary at a premium. On the

trade.xyz perp the order was inverted: the wallets that would not have received a share allocation captured the most upside on the only continuous market that priced the deal in real time.

Conclusion

Traditional financial markets are full of products the average retail consumer cannot reach. Pre-IPO exposure is the cleanest illustration of that asymmetry. Oversubscribed IPOs clear overwhelmingly to clients of the underwriting banks, the secondary market opens at whatever price retail demand can sustain against the institutional cost basis, and the venues that exist to bridge that gap (SPV platforms, forward contracts, prime-broker derivatives) replicate the access conditions of the primary market rather than dissolve them.

Hyperliquid is moving in a direction that flattens that gradient. The IPOP design is one of several primitives the venue is shipping to bring instruments that have historically been institutional-only into a retail-accessible perpetual form. A retail account on a HIP-3 deployer pays the same fee schedule as a hedge fund on the same contract, and both fill against the same book. The thesis the project keeps coming back to is that Hyperliquid intends to House all Finance, and the CBRS perp is the first hot US tech IPO that ran through that infrastructure end-to-end.

Cerebras priced on May 13 at $185 above its raised range with books reportedly twenty times oversubscribed. The

trade.xyz

CBRS perpetual cleared $70.7M of pre-pricing notional across 208,854 fills and 2,148 distinct addresses over twelve days. It tracked the underwriter's range through every meaningful event, absorbed institutional-scale conviction without breaking its quote, and converged toward its post-conversion mark within minutes of the pricing wire. The first major US tech IPO with a continuous pre-IPO perpetual on a Hyperliquid deployer happened twelve days ago. There are more coming. Welcome to Pre-IPO summer.

Hyperliquid.

This article is published for informational and educational purposes only and does not constitute investment advice. Arrakis has made reasonable efforts to verify the accuracy of the data presented but does not warrant that all information is accurate, complete, or current.

Written by:

*@web3_pastel*

Designed and Edited by:

*@oxillustration*

and

*@mchammond*