How Trade[XYZ] Built 92 Markets on Hyperliquid and Reached 98% of HIP-3 Volume

Author: Mohit Pandit

Translated by: TechFlow

With data current through June 2026, the debate around Trade[XYZ] and Hyperliquid has turned sharper. As HIP-3 open interest has climbed, volume share has expanded, and more pre-IPO assets have gone live, one question keeps resurfacing: is Trade[XYZ] becoming a threat to Hyperliquid itself, and by extension to HYPE?

This analysis argues the opposite. Its conclusion is that Trade[XYZ] is not weakening Hyperliquid. It is showing that Hyperliquid’s design works: the base layer provides neutral, high-performance infrastructure, and specialized operators compete on top of it to build markets, with liquidity deciding the winner.

92 markets in eight months, about 98% of HIP-3 volume

The report says Trade[XYZ] built perpetual markets for equities, indexes, commodities and foreign exchange in eight months and now accounts for roughly 98% of HIP-3 trading volume. In the author’s view, that is evidence that HIP-3 can already support non-crypto perpetual verticals built by professional teams and backed by institution-grade liquidity.

In that setup, Hyperliquid still keeps the users, matching engine activity, fee sharing, auction demand and ecosystem narrative, while avoiding direct listing execution and regulatory exposure. The author says the usual argument is too narrow: yes, Trade[XYZ] locks HYPE, launches and runs new markets, generates fees, and channels part of the economics back toward HYPE buybacks. But that understates the relationship between the deployer and Hyperliquid.

Two ways to build a derivatives exchange

The piece splits exchange expansion into two paths.

• A vertical path, where the venue builds every market itself, sources assets, runs oracles, recruits market makers, takes on the risk and keeps the upside. Lighter and Ostium are cited as vertically integrated products, with Ostium described as a pure RWA venue.
• A horizontal path, where the base layer is provided and permissionless deployers build markets on top of it while sharing fees with the underlying exchange. Hyperliquid’s HIP-3 is presented as that model, and @tradexyz as one of its deployers.

The author says HIP-3 should not be read as horizontalization for its own sake. A better frame is that it acts like an admission mechanism. Hyperliquid is making a bet that the most durable edge in onchain finance sits in core infrastructure: the L1, the clearing layer and the matching engine. If that infrastructure is strong and neutral enough, the best operators will choose to build on it.

The report ties that idea to how financial markets usually develop. There is one CME, one NYSE, one HKEX. Liquidity pulls in more liquidity. In a category where no single venue wins deep liquidity, the category has effectively already lost. From that angle, Hyperliquid is trying to become the home venue for all finance, with HIP-3 serving as the mechanism that lets different operators compete to build category-winning markets on top of a shared base.

The two main objections

The report lays out the pushback directly.

The first argument is that Hyperliquid is giving away future value. The deployer keeps roughly half the fees and has a kind of franchise position, so Hyperliquid is said to be surrendering revenue it could have captured by building equity perpetuals itself.

The second objection is more pointed. HIP-3 is supposed to look open, but one deployer now drives about 98% of the volume. That has led to accusations of favoritism, often tied to Trade[XYZ]’s links to the Unit ecosystem, while Hyperliquid still collects 50% of exchange fees.

The author’s answer is that both criticisms understate how hard it is to build a real institution-grade RWA market. The purpose of the piece is to test, with data and first-principles reasoning, whether the current model has shown any real sign of success. In the author’s reading, it has.

Listing is the easy part. Tradeable depth is the moat.

One of the report’s central points is that “just list the asset” badly misreads the business. Putting a new ticker on the screen is easy. The hard part, and the moat, is making that market trade in meaningful size.

The Trade[XYZ] data is used to highlight three problems that have to be solved:

1. Launch quickly enough to catch demand.
2. Bring in market makers that can create real depth.
3. Keep that liquidity economically real while running the markets day after day.

Launch speed

Measured onchain from asset registration to first trade, the median time to launch was 3.3 days. The report says 65% of markets went live within a week and 47% within three days.

That matters because a perpetual market only has value if it already exists when traders want exposure.

Depth, not just breadth

The report presents market depth figures to argue that the books are not merely populated but genuinely tradeable.

• XYZ100 had $2.6 million of resting orders within 10 basis points of mid.
• The S&P 500 market had $964,000 in the same band.
• Gold had $759,000.
• Single-name books such as NVIDIA and Tesla were said to offer comfortable size as well.
• By contrast, the median market had only about $20,000 within 10 basis points.

The interpretation is straightforward: rational market makers do not spread capital evenly. Flagship markets get institution-grade depth, while long-tail markets stay much thinner.

Maker counts, spreads and volume move together

Across 73 markets with sufficient data, the report gives three correlations:

• Daily distinct market-maker wallets versus spread: -0.72.
• Volume versus spread: -0.82.
• Volume versus open interest: +0.96.

The volume-weighted average spread across the books was 2.33 basis points, while daily turnover ran at about 2.9x open interest. The author argues that Trade[XYZ]’s real edge is not the listing event itself but the work of getting market makers in the door and deploying capital in a way that produces tight, durable books.

Why equity perps are harder than crypto perps

The report then steps through the market-making problem from first principles.

Market makers live on spread capture, but only if they can manage what gets left on their books after each fill. In crypto perpetuals, a desk can hedge around the clock on another crypto venue. In equity perpetuals, the real hedge is usually the underlying stock, ETF or futures contract, and those only trade while the relevant cash market is open.

That changes the risk profile immediately. During regular hours, a desk can hedge a TSLA perp inventory with TSLA stock and quote tight, deep prices with little incremental risk. After the market closes, the desk is left warehousing naked inventory. The rational response is wider spreads, less depth or no quoting at all. For pre-IPO assets, the situation is even more severe because no practical hedge exists, which is why those books tend to be thin before listing.

The author also points to other frictions:

• Adverse selection, since a larger share of after-hours flow may be informed.
• Funding and carry, because funding must keep perps anchored without making hedging uneconomic.
• Oracle and gap risk, since stale, manipulable or discontinuous marks create liquidation risk that market makers cannot control.

Discovery bounds, liquidation protection and funding multipliers

The report argues that Trade[XYZ] made these markets workable through a toolbox of risk controls rather than through any single feature.

Discovery bounds keep the mark price within plus or minus one turn of the maximum leverage versus the reference price. At 20x leverage, the report says that works out to about 5%. The boundary re-anchors in discrete, market-capped steps and remains a hard limit until outside pricing returns.

That is paired with liquidation protection. If a position’s liquidation price falls outside the active boundary, the position is not liquidated. In practical terms, the author says this gives desks a known ceiling on how far price can move in one step during unhedgeable periods, turning the worst overnight inventory outcome into something bounded and quantifiable rather than open-ended.

Funding settings are adjusted as well. On a per-market basis, the standard funding rate is scaled by 0.5, which the report says is roughly a 5.5% annualized baseline, while pre-IPO products are cut to 0.005. The goal is to keep perps tied to fair value without crushing market makers with carry costs, especially in products where no stock exists to arbitrage against. Taken together, the author sees this as a toolkit for markets that, by first principles, should be nearly impossible to quote once hedging disappears.

Overnight depth held up, though not without limits

The report measures order-book depth across the top 10 equity books at different times of day and finds two notable patterns.

First, overnight depth held at about 116% of cash-session levels. In names such as NVIDIA and Tesla, depth actually increased after hours because the perp became the only active price venue once the stock market closed, concentrating quotes there.

Second, weekend depth fell to about 37% of normal when even index futures were shut and hedges disappeared for two full days.

The author does not present that as some magical after-hours advantage. The point is narrower. Trade[XYZ]’s risk controls appear to keep makers quoting overnight in books that first-principles reasoning would suggest should collapse. The durable edge still lies in daytime depth, order flow and breadth across products that are genuinely hard to make markets in.

This is not a one-time listing business

Trade[XYZ], the report says, did not simply launch markets and walk away. Over roughly the latest 300 onchain action windows, it carried out 294 distinct risk-management operations:

• 54 position-limit changes.
• 35 growth-mode switches.
• 34 funding-multiplier adjustments.
• 28 trading halts.
• 11 margin-mode changes.
• Plus asset-level tagging.

That is ongoing market-by-market risk work across 92 underlyings, dealing with real trading sessions, halts and funding settings. The author’s description is blunt: this is a full-time market-operations business.

Compared with other onchain equity and RWA efforts, Trade[XYZ] is ahead

The report uses a set of comparisons to frame the scale.

xStocks, tokenized spot equities on Solana, represent more than $25 billion in total trading volume, but actual DEX volume is only about $517 million. Ostium, a dedicated and funded RWA perpetual DEX, has done about $59 billion in cumulative volume, yet its open interest stands at only about $115 million, or roughly one twenty-fourth of Trade[XYZ].

Newer entrants such as Variational are described as taking a different approach altogether, not trying to build native depth but instead aggregating liquidity by RFQ from Hyperliquid, Lighter and centralized exchanges, then routing into Hyperliquid to access the liquidity under discussion.

The author’s conclusion is that the leading onchain equity-perpetual venue is not a separate standalone exchange. It is Trade[XYZ] on Hyperliquid, and by a wide margin.

About 97% of activity still runs through Hyperliquid’s own app and API

A natural assumption is that deployers own the user relationship through their own front ends. The report says the data shows the opposite.

Each trade was tagged by the builder code that generated the front end behind it, measured on the taker side, meaning the side that chose the interface. The result: about 97% of Trade[XYZ] market volume was traded through Hyperliquid’s own app and API. All third-party front ends together accounted for about 3%, and Trade[XYZ]’s own front end was only a small slice of that.

In other words, almost every trade in these markets is still happening inside Hyperliquid’s interface rather than being siphoned away from it.

More than 300,000 wallets acquired for Hyperliquid

The report says Trade[XYZ] has brought in more than 300,000 distinct wallets to Hyperliquid. Monthly additions are still running at roughly 36,000 to 48,000, and peaked near 79,000 in March during the launch wave and the surge around SpaceX.

The author frames equity and RWA perpetuals as top-of-funnel acquisition products. The assets draw attention, but Hyperliquid is where users land, trade and remain. That user-acquisition and attention value, the report says, will never fully show up in a fee table, even though it is a real part of the network effect.

Fee split: about $37.9 million paid so far

The protocol-level incentive structure is broken down in more detail.

HIP-3 traders have paid about $37.9 million in fees. The report says that amount splits three ways:

• Roughly $9.2 million went to builder-code fees paid to third-party front ends, not to the deployer.
• The remaining exchange fees are split 50/50 between Hyperliquid and the deployer.
• Hyperliquid’s protocol share comes to about $14.3 million and is directed toward HYPE buybacks, while another roughly $14.3 million accrues to the deployer.

The piece adds that HIP-3 caps the deployer take. If the deployer share exceeds 100%, Hyperliquid’s protocol fee matches the excess, meaning the deployer can never collect more than half of the exchange fee.

Growth mode cuts fees by about 90%, but it is not the whole story

HIP-3 lets deployers choose a fee mode for each market.

In standard mode, takers pay 9 basis points and makers pay 3 basis points. In growth mode, those fall to 0.9 and 0.3 basis points, a reduction of about 90%. According to the report, growth mode is limited to non-crypto real-world assets and explicitly excludes crypto wrappers such as MSTR. GOLD is also excluded because it overlaps with the existing PAXG-USDC market.

The author treats that exclusion as a natural experiment. Today, books that qualify for growth mode are charging around 0.86 basis points, while excluded products are around 7 basis points, an 8x gap on the same matching engine.

The argument is that RWA perpetuals compete on total cost with traditional finance. Charging 9 basis points does not work against CME index futures or stock commissions. Charging 0.9 basis points is far more competitive, while still offering 24/7 access and leverage.

Three data points against the “low fees create volume” argument

The report still argues that cheap fees alone do not explain the volume. It gives three reasons.

First, the onchain control group. Six of the other seven HIP-3 deployers have access to the same fee tool but almost no volume. The second-ranked deployer, dreamcash, even quotes tighter spreads yet remains around 30 times smaller. If low fees were enough by themselves, the gap should not be that wide.

Second, GOLD works as the counterexample. It pays about 8 times the fee of growth-mode books, yet it is the single largest fee market and still ranks in the top three by both volume and open interest. Traders, in the author’s view, are willing to pay full fees for GOLD because the liquidity is there.

Third, that leads into a monetization argument. Switching off growth mode would not necessarily kill volume, but it would route more value to HYPE. Because exchange fees are split 50/50 in both modes, raising fees would increase the value flowing toward HYPE by about 9x to 15x, from roughly 0.9 basis points in growth mode to about 9 to 12 basis points in standard mode. Even with a large drop in volume, Hyperliquid’s buyback share would still rise unless volume collapsed by more than about 85%.

At the 7-basis-point level observed in GOLD, the report says tradexyz would need only about 11% of its current volume to match today’s buyback output. At 5 basis points, that threshold is about 15%. At 3 basis points, about 25%. The author sketches one possible monetization path: move mature markets to 5 to 7 basis points and, assuming half to three-quarters of volume is retained, direct roughly $90 million to $185 million a year to buybacks, or about 3x to 5x the current level.

The report says this is not just a hypothetical. GOLD already runs at standard fees, converting 4.3% of volume into 23% of all buybacks. That is presented as live evidence that deep RWA markets can continue trading at standard rates and that a volume collapse greater than 85% is unlikely.

Top 30 markets hold about 95% of open interest

On market structure, the top 30 books account for about 95% of open interest, led by the S&P 500, the XYZ100 index, Brent crude and WTI.

The more interesting pattern, the author argues, is how quickly each market got there. Measured from listing to the point where a market reached 25%, 50% and 75% of its current open interest, the median market took 9 days to reach one quarter of its eventual size, 15 days to reach half, and 30 days to reach three quarters.

The spread between markets is wide. The fastest books reached half their current open interest in about two weeks: SpaceX took 14 days, while the S&P 500 and silver were about 15 days. Early single-name listings, by contrast, were launched when the venue’s liquidity infrastructure was still immature and took five to six months: Microsoft 192 days and Meta 159 days.

The author reads that as a visible learning curve. More recent listings are growing much faster than the early batches because market-maker relationships and risk tooling now exist from day one.

Maker concentration and market quality

Top-five concentration has eased in flagship books

A heatmap in the report tracks the share of passive volume controlled by the top five market makers in each market over time. Early markets showed deep-blue concentration, with a few firms supplying nearly all passive liquidity and top-five shares above 90% in the first months.

Over time, the largest and most liquid markets lightened in color as more makers competed at the top of the book, while many single-name equities remained concentrated. The report does not frame concentration itself as a flaw. Market seeding often starts that way. But it says the move toward competition in flagship books is a healthy sign that liquidity provision on tradexyz is now a competitive business rather than a favor from one or two makers.

A small set of maker wallets appears across many books

The report also ranks the top passive participants in each market over the past 30 days and checks which wallets repeatedly show up across names. The result points to a clear core:

• The single biggest anchor wallet is a top-five maker in 47 of 73 markets.
• It ranks number one in 22 of them.
• The top three anchor wallets together place in the top three of 57 out of 73 markets.

Some of those wallets quote across all four asset classes: equities, commodities, FX and indexes. All are described as having textbook market-maker profiles, with directional exposure inside 1% and realized PnL close to zero.

Fee generation is led by commodities and indexes

The fee base is not evenly distributed. Commodities account for 54% of all earned fees, indexes for 24%, and the combined long tail of single-name equities and FX for 22%, even though equities make up most listings.

By individual market, gold is the largest fee contributor at 23%, or $8.7 million. It is followed by the XYZ100 index at 18%, WTI crude at 13%, and silver at 10%. The top 10 markets generate 84% of all fees.

The report stresses a nuance that GOLD makes impossible to ignore: fee rank is not the same as volume rank, because fee mode differs by market. GOLD is the only major market excluded from growth mode, so it pays around 7 basis points while most of the rest of the order book pays around 1 basis point. That alone makes it the number-one fee market. It contributes just 4.3% of volume but 23% of all fees. By trading activity it is secondary. By buyback fuel it is huge.

Could Hyperliquid’s core team have built this itself?

The report’s answer is no, and more importantly, it says they should not.

The strongest reason is regulation. Listing perpetuals on NVIDIA, TSLA and pre-IPO SpaceX sits squarely in securities-derivatives territory. HIP-3, in the author’s reading, is deliberately designed to externalize that responsibility to deployers. If the core team listed those markets directly, the protocol, the foundation and HYPE would move much closer to regulators. Keeping listings at arm’s length is not a missed opportunity in this framing. It is the design.

The rest of the reasoning builds from there:

• Hyperliquid’s value rests on being neutral infrastructure, and core-team asset selection would weaken the permissionless case.
• HIP-3 is also meant to monetize a deployer-auction fee market, and internalizing listings would disrupt that mechanism.
• Running 92 equity, FX and commodity markets, sourcing oracles, handling market hours and halts, cultivating makers and carrying out hundreds of visible onchain risk operations is a full operating business, separate from building a high-performance exchange.
• Winning top-tier makers in niche RWA perps is relationship and capital work, not protocol engineering.

The report points to the market record as supporting evidence. If the task were easy, or if it could be done in-house without tradeoffs, the ecosystem would likely already show multiple strong deployers or a direct effort from the core team. Instead, the second-largest deployer is said to be 46 times smaller, dedicated RWA venues are 24 to 33 times shallower, and new entrants are routing liquidity back into Hyperliquid.

The article closes with an analogy the author says had the biggest impact on his thinking: Tether opened global access to dollars, and Trade[XYZ] is doing something similar for global access to global equities. It also notes that all data used in the piece was provided by the team behind @hydromancerxyz.