Author: Lilian, COO and co-founder of OAK Research; Translation: @金财经xz

In just two years, Hyperliquid has proven that it is still possible to innovate, build and distribute a truly usable on-chain financial product that attracts users, generates revenue and competes with centralized giants. Although it sounds simple, this is still a major breakthrough for the industry.

Early last year, we released research on Hyperliquid’s vision, describing it as the “AWS of Liquidity.” Just as AWS revolutionized cloud infrastructure by abstracting away technical complexity and letting developers focus on building products, Hyperliquid has done the same for liquidity.

Hyperliquid’s true success lies not just in its product quality, execution efficiency or token economic model. Ultimately, the key is one thing: building a high-performance infrastructure that attracts and retains liquidity. Hyperliquid’s true competitive advantage lies in its ability to transform liquidity into a resource accessible to everyone.

Every stage of Hyperliquid’s development confirms this. Builder Codes is based on a simple concept: instead of acquiring users through traditional channels, product distribution is outsourced to any third-party application that wants to provide trading services to its users by connecting to the Hyperliquid infrastructure.

Following the same logic, HIP‑1 introduced native spot assets and their supporting order books. HIP‑2 creates a liquidity guidance mechanism. HIP‑3 materializes Hyperliquid’s core philosophy—allowing external deployers to create perpetual contract markets without permission while leveraging the most performant and liquid infrastructure on the market.

HIP‑4 follows exactly the same path. It introduces a new class of market that is still relatively unknown in on-chain finance, brings an entirely new form of revenue, and opens up a new playing field for deployers and builders - they will be responsible for distribution and monetization, while Hyperliquid continues to monetize its infrastructure in the background.

Today, traditional financial markets are largely dominated by products with non-linear returns: options, credit default swaps (CDS), insurance products, structured products, etc. Most of these markets are either absent or have only very few representatives in on-chain finance.

HIP‑4 is essentially Hyperliquid’s attempt to fill this gap: integrating outcome markets directly into a financial infrastructure that already supports spot and perpetual contract markets, while reusing the same deployment logic that made the previous HIP series successful.

1. What is HIP‑4?

(1) Overview

HIP‑4 is a new primitive introduced by Hyperliquid called "Result Market". These are fully-collateralized financial contracts that settle within a predefined price range, with the final value dependent on whether a specific event occurs or not. They can be compared to prediction markets, or tools similar to bounded options.

An outcome market operates according to a simple structure: define a proposition about a real-world event, set an expiration date, specify settlement conditions, and then the market gradually converges to a final value between 0 and 1.

In its simplest form, the outcome market behaves as a binary market. There are two positions: YES and NO. If the event occurs, the YES token will be settled with 1 USDC and the NO token will be settled with 0; if the event does not occur, it will be the opposite. The contracts trade freely on two separate order books before final settlement, but remain linked through negate and merge operations, allowing them to share liquidity.

The prices of YES and NO tokens directly reflect the market’s implicit expectations of the probability of this event occurring. If the YES token trades at P, the NO token mechanically trades at 1‑P. This mechanism ensures that the two results are always mathematically consistent (YES + NO = 1).

The first HIP‑4 marketplace that has been deployed on Hyperliquid clearly demonstrates this mechanism. The proposition is: "Will the closing price of BTC be higher than ___ USD tomorrow at 06:00 UTC?" A trader who buys YES for 0.62 USDC is essentially paying 62 cents for the event to occur in exchange for a potential $1 gain. If the closing price of BTC is lower than the specified price, the contract will return to zero, and the maximum loss will be the principal invested.

At first glance, the structure looks quite similar to traditional prediction markets such as Polymarket or Kalshi. However, the fundamental difference is this: HIP‑4 Results Marketplace is not designed as a consumer-facing application and its goal is not to provide the most viral marketplace possible and the most intuitive front-end interface. In fact, they are designed to provide Hyperliquid with a new category of financial instruments to complement its existing product line.

(2) Results Market Structure

From an architectural perspective, HIP‑4 is natively integrated into HyperCore. This means that the resulting market shares the same CLOB (central limit order book) infrastructure, matching engine, order types and performance as the spot and perpetual contract markets. We are talking about a theoretical processing capacity of 200,000 orders per second, with settlement finality completed within one block. No separate systems, no parallel infrastructure, and no migrations required.

Specifically, the market is created through an OutcomeDeploy request, defining several key parameters: settlement conditions, expiration date, settlement data source, collateral assets, and broadcasters with the authority to publish the final market results. This architecture is directly inherited from the basic framework of HIP‑3.

This means that HIP‑3 deployers can use the same deployment slots and 500,000 HYPE pledge quota to launch the result market. The result module is just an additional parameter in the request: including this parameter creates a result market, omitting it creates a standard perpetual contract market. Deployers who are already active in the HIP‑3 ecosystem can seamlessly access HIP‑4 without any additional steps.

HIP‑4 The life cycle of the market is divided into four stages:

Phase 1: Market Deployment - Define all parameters.

The second phase: Auction starts - lasts about 15 minutes, uses a single price auction mechanism, collects orders before immediate execution and settles at the equilibrium price. This stage is critical for initial price discovery in new markets with no trading history.

Phase 3: Continuous Trading — Continuous trading on the shared YES/NO order book until expiration. It is worth noting that the transition from the second phase to the third phase is achieved by migrating the remaining auction orders to the continuous order book.

Phase 4: Settlement — The authorized broadcaster publishes the final value, the protocol automatically settles the position, and USDC is distributed to the winning holder.

The document also introduces the concept of "Questions". These are outcome markets that contain multiple outcomes, where exactly one outcome settles as YES and the rest all settle as NO. This mechanism supports polynomial markets such as: "What will be the US inflation rate next June?" with multiple possible outcomes (<3.3%, <3.4%, <3.5%, etc.).

The merge and negate mechanisms between order books avoid unnecessary dispersion of liquidity between independent markets while maintaining a consistent relationship between implied probabilities.

(3) Other points

Risk Management, Collateral and Settlement:

Prior to HIP‑4, Hyperliquid used a continuous risk engine designed specifically for the perpetual contract market. As a result the market follows a completely different logic: since returns are bounded from the outset, the maximum risk of a single position is determined when the trade is executed. A trader who buys the YES contract at 0.62 USDC will never have a maximum loss of more than 0.62 USDC, and a maximum profit of 0.38 USDC.

This feature may seem simple, but it has brought about major changes in risk management and settlement mechanisms. HIP‑4 positions are fully collateralized upon creation, eliminating a lot of complexity. There is no need for an aggressive liquidation (liquidation) engine as the protocol already holds the maximum potential losses on both sides of the market.

Discontinuous Payoffs:

This is the key feature mentioned earlier that allows Hyperliquid to introduce great flexibility in revenue design. Returns in the spot and perpetual contract markets are linear (the higher the underlying asset price, the position return increases linearly), while HIP‑4 introduces a wide range of nonlinear returns (binary, convexity, concavity, intervals, barriers, path dependence, etc.), which we will explore in detail in the subsequent sections of this article.

HIP‑4 Deployment Roadmap:

We are currently in the first phase of the HIP‑4 market. This phase is intentionally controlled, limited to the “canonical” market launched by the Hyperliquid team. The goal is to ensure that there are no issues with yield, settlement or condition determination in the early market.

Phase 2, expected soon, will open permissionless deployment to all builders. It will introduce a slot recycling mechanism: a single 1 million HYPE pledge can support a series of markets rolling over time. This will incentivize deployers to create cyclical markets (such as daily BTC markets, earnings releases, Fed meetings, etc.) rather than temporary viral markets.

Aggressive fee structure:

Opening or minting HIP‑4 positions is free. Fees follow HyperCore's spot rate table (7 bps for takers/4 bps for makers) and are only charged when closing or settling a position. The SetOutcomeFeeScale parameter allows adjusting the fee specifically for prediction markets, but is currently set to zero in canonical markets.

Compared to Polymarket’s fees of up to 2% on winning positions, or Kalshi’s tiered fee model for active traders, the current difference is extremely advantageous.

Portfolio Margin Integration:

HIP‑4 positions are integrated directly into Hyperliquid’s portfolio margining system. They coexist in the same account as spot and perpetual positions and share the same USDC collateral. The trading volume generated by the resulting contract also counts toward the account's overall rate tier.

2. Three revenue dimensions of HIP‑4

Most of the current commentary surrounding HIP‑4 still views its outcome market simply as an extension of traditional prediction markets. This interpretation is relatively one-sided as it confuses visible use cases with the actual financial primitives Hyperliquid is introducing.

Hyperliquid explicitly describes HIP‑4 as suitable for “applications in prediction markets and similar bounded options instruments.” The second half of this sentence is the most critical. While prediction markets typically aim to create markets around viral events to drive speculation, outcome markets aim to introduce an entirely new yield primitive to Hyperliquid.

A HIP‑4 Outcome Contract can be configured in three independent dimensions, the combination of which defines the final product. Let’s break down its meaning one by one.

(1) Settlement Input

The first dimension is the object of market measurement. It can be an asset price, interest rate, spread between two assets, on-chain metrics (like TVL or protocol revenue), protocol status (exploits, decoupling, upgrades), macroeconomic data, or any external data verified by an authorized broadcaster.

This is not limited to crypto asset prices, nor is it limited to public-facing events. A market can settle based on the perpetual contract market’s average funding rate over the week, the proportion of liquidations triggered during a certain trading session, or a company’s earnings per share. Anything that can be objectively observed and verified is a valid settlement input.

This flexibility is fundamental because it transforms HIP‑4 into a risk characterization infrastructure rather than just another directional market. As long as data can be objectively defined and parsed through an identifiable settlement source, it has the potential to be used in outcome markets.

(2) Observation Mode

The second dimension defines how the input value is measured. The value can be observed at the exact expiration moment, within a time window, or based on the maximum/minimum value reached during the period, or based on the duration for which the threshold was exceeded, or simply whether the threshold was touched during the window period.

This dimension greatly expands application scenarios. For example, "Will BTC be above $100,000 at expiration?" and "Has BTC ever touched $100,000 during the week?" are two completely different products with different risk characteristics, probability distributions, and trader uses.

In the first case, only the final value matters; in the second case, the market becomes path‑dependent, i.e. just crossing the threshold at any moment is enough to trigger a return.

However, both products are built on the same primitives, use the same settlement inputs, and differ only in the observation mode. This is the strength of HIP‑4: the ability to open the infrastructure to a whole class of traditional derivatives, such as barrier options, range options and certain exotic structured products.

(3) Payoff Shape

The third dimension is the most critical: the revenue function itself. It determines how observed outcomes translate into economic value for contract holders.

In the first HIP‑4 markets, this function is binary. The contract is settled to 0 or 1 depending on whether the event occurs. However, this is only a very basic form of discontinuous returns.

Recall that spot and perpetual contracts rely on linear returns because their returns evolve linearly relative to changes in the price of the underlying asset. In contrast, the financial field contains a wide variety of nonlinear returns: binary, convex, concave, interval, barrier, path dependent, volatility-linked, etc.

HIP‑4 introduces the possibility of integrating new revenue structures into Hyperliquid. One can imagine scalar payoffs, where the final value is a continuous function of the observed outcome; range-based payoffs, where the payoff is maximized if the result falls within a preset range; or composite payoffs, which are built on multiple branches.

With HIP‑3, deployers gain market creation rights (deciding which markets are tradable). With HIP‑4, they now have benefit structuring rights (deciding how claims are settled). This is no longer just a new asset, but a new form of tradable risk.

3. Why does Hyperliquid need HIP‑4?

(1) The demand is huge, but there is no real supply in the chain

Before explaining how HIP‑4 will structurally change prediction markets, it is necessary to start with a simple observation: financial instruments with non-linear returns account for a huge proportion of traditional financial markets, but are still extremely underdeveloped in on-chain finance.

The options market is the most obvious example. The Cboe report shows that 2025 is the sixth consecutive year that U.S. listed options have set trading records. The total trading volume for the year exceeded 15.2 billion contracts, and the average daily trading volume reached 61 million contracts. "Zero-day expiration options" (0DTE) alone account for 59% of SPX options trading volume, with an average daily trading volume of 2.3 million contracts.

In the crypto field, Deribit dominates the options market, with a trading volume of US$1.875 trillion in 2025 and a market share of over 85%. It is a centralized, off-chain platform that is structurally disconnected from on-chain liquidity. Prediction market volume will reach $63.5 billion in 2025, compared with $15.8 billion in 2024, and the annualized run rate in 2026 will be approximately $250 billion.

All of these markets have one fundamental thing in common: they allow users to express views on the future state of the world, rather than just directional exposure to asset prices. Although the success of traditional finance proves that demand is real, these markets are either missing on the chain or highly fragmented.

This is the missing key puzzle piece of on-chain finance. At least until HIP‑4 came along. Therefore, we believe that HIP‑4 is less about Hyperliquid’s expansion into prediction markets and more about solving a major unmet need among traders.

(2) Main limitations of perpetual contracts

Prior to HIP‑4, HyperCore essentially only offered two main forms of exposure. The spot market allows users to hold assets directly, with returns linearly related to price changes. Perpetual contracts provide leveraged directional exposure, but their return structure is also fundamentally linear.

For a perpetual market to function properly, three inseparable components are needed: a continuous oracle that tracks the price of the underlying asset in real time, a funding rate mechanism that maintains the convergence of perpetual prices and spot prices, and a liquidation engine that manages leveraged positions. This architecture works well when the underlying asset is highly liquid, has a continuous price, and is readily observable.

However, once applied to discrete events, this structure appears to be a gross mismatch. For example, the proposition "Is the closing price of BTC higher than $100,000 this Friday?" is not an asset with a continuous price curve. As expiration approaches, market prices no longer track a continuously evolving underlying asset, but instead track a probability that must suddenly converge to 0 or 1. Currently, there is no natural continuous oracle for this type of market.

This discontinuity makes market making significantly more difficult, increases the risk of arbitrage between order books, and may even lead to absurd situations where fully solvent positions become unmanageable in an engine designed for a continuous pricing environment. In other words: This structure cannot be shoehorned into a perpetuity model.

HIP‑4 is also a technical solution to reality: Perpetual markets and HIP‑3 were not designed to support financial instruments with non-linear returns at all. Once again, HIP‑4 should be understood as an extension of Hyperliquid’s capabilities rather than a simple addition to the market.

4. The changes HIP‑4 brings to the prediction market

(1) Next step: risk market

Naturally, HIP‑4 Market is often compared to the products offered by Polymarket or Kalshi. In fact, we believe HIP‑4 represents the next stage in the natural evolution cycle of prediction markets. Specifically:

The first phase in the history of prediction markets was the era of “information markets”: tools designed to aggregate information to generate probabilities more efficiently than polls, expert opinions, or traditional forecasting models.

The second phase is the era of "speculative markets": the phase we are currently in, where users turn opinions into bets on real-world events through simple, intuitive, speculation-oriented platforms.

Polymarket perfectly embodies this stage. The platform proves that prediction markets can become a consumer-facing product that engages users, generates valuable signals, and becomes a reference point for predicting major events such as the U.S. presidential election. But this success does not mean that the current form is the final form of the field.

The third stage is the era of "risk markets": in this stage, the market no longer exists purely for prediction or betting, but becomes a tool to create new financial exposures for specific risks. The goal is to use real-world events as a new way to trade - both for directional trading and as a hedging tool.

This is exactly what HIP‑4 achieves. Rather than building a better prediction market interface than existing products, it integrates outcome contracts into the same underlying infrastructure as perpetual contracts, spot markets, and HIP‑3 markets. For the first time, traders can view outcome markets as part of an overall portfolio strategy rather than as an isolated bet.

This distinction is crucial. The traditional prediction market is a place where users go to speculate on an event; the "risk market" is a tool integrated into portfolio construction and risk management strategies. HIP‑4 is clearly pushing the outcome market into this second category.

(2) Unified margin is a fundamental breakthrough

One of the most interesting and probably severely underrated aspects of HIP‑4 is its integration with the Hyperliquid portfolio margining system. However, it is crucial to understand why outcome markets can become so much more than simple prediction markets.

Portfolio Margin unifies spot and perpetual accounts to improve capital efficiency, while also allowing eligible idle assets to generate income through the "earn" function (mainly USDC for traders).

Polymarket uses an isolated positions model: funds invested in a certain market are completely isolated from the rest of the user's investment portfolio. Market A positions cannot be used as collateral for Market B positions. To make matters worse, users cannot hedge positions on Polymarket with perpetual positions elsewhere.

Hyperliquid solves this problem because a HIP‑4 outcome contract can coexist with spot and perpetual positions in the same account. Therefore, the same collateral can simultaneously support a long BTC position, as well as a NO position targeting “will BTC exceed $___ tomorrow?” This significantly improves capital efficiency for both traders and market makers.

This last point is extremely important because the technical structure of platforms like Polymarket or Kalshi makes it difficult for market makers to operate efficiently. On Hyperliquid, market makers can easily hedge HIP‑4’s market making strategy through spot or perpetual positions.

Additionally, this opens up huge possibilities for builders - they can now create structured products that combine directional perpetual exposure with conditional returns within a single interface. This is exactly what is needed in the “risk markets” phase: tools that can be integrated into broader portfolio logic, collateral efficiencies and dynamic risk management systems.

(3) Fewer markets, more precise definitions, higher liquidity

Another major difference between HIP‑4 and current prediction markets is the logic behind market creation and curation. Platforms like Polymarket and Kalshi benefit from listing on as many markets as possible to maximize engagement, virality and social media exposure. This works perfectly for a business model centered around speculation and consumer entertainment.

Hyperliquid has no incentive to replicate this model. The economic mechanism of HIP‑4 will naturally drive the market toward cyclicality, standardization, and high liquidity, thus attracting traders. In other words, the logical outcome should be the formation of an ecosystem with fewer markets but greater integration with existing tradable assets, since the goal is to transform the resulting markets into actual financial instruments.

Additionally, slot recycling mechanisms may incentivize deployers to optimize single slots around high-potential cyclical markets, rather than spreading liquidity across low-liquidity long-tail markets. An illiquid market that cannot scale is not a useful risk management tool; a cyclical market with clear settlement, sufficient depth, and native integration into margin accounts can be such a tool.

(4) Advantages of cold start

Prediction markets face the core problem that plagues most on-chain finance: liquidity. Without liquidity, there are no market makers, without market makers, there is no trading volume, without trading volume, there are no users, and without users, there is no liquidity. It was a vicious cycle that took Polymarket several years to finally break, thanks in large part to the major catalyst of the U.S. presidential election.

HIP‑4 has a huge advantage: a ready base of active traders, professional market makers, deep liquidity, and ultra-high-performance infrastructure. Market makers, builders and deployers are already connected to Hyperliquid to distribute HIP‑4 markets without rebuilding the entire trading infrastructure from scratch.

This does not guarantee that HIP‑4 will succeed, but it fundamentally changes the typical "cold start" problem faced by most protocols. This is probably one of Hyperliquid’s biggest advantages: it doesn’t need to bootstrap liquidity from scratch.

(5) Infrastructure and consumer business

It is inevitable to compare Hyperliquid with Polymarket and Kalshi, but this comparison is often based on the wrong framework. Polymarket and Kalshi are essentially "consumer-facing distribution businesses." Their advantages come from brand, user experience, regulatory positioning, media attention and, in Kalshi's case, a clearer regulatory framework in some U.S. markets.

Hyperliquid follows a completely different logic. HyperCore does not necessarily attempt to be the best prediction market interface for retail users. Its value proposition is to provide execution, margin and liquidity infrastructure upon which other participants can build products.

This is why lower fees alone may not be enough to directly attract Polymarket’s retail user traffic. A user betting $50 on the outcome of an election would not necessarily choose a platform based on a fee difference of a few basis points. But for active traders, market makers or builders, the difference between expensive, isolated, non-composable infrastructure and native, highly liquid, deeply integrated infrastructure becomes extremely meaningful.

As this section should make clear, the primary value of HIP‑4 is in creating a new transaction primitive for users, improving capital efficiency, and opening up new monetization opportunities for builders. This is exactly the logic behind our “Liquid AWS” argument around the Hyperliquid vision.

5. Risks faced by HIP‑4 and Hyperliquid

(1) Settlement risk

The first major risk surrounding HIP‑4 is obvious: the reliability of an outcome market depends entirely on its settlement process. In a market like “Is the BTC price above $100,000 at 06:00 UTC tomorrow?” settlement is relatively straightforward. The data sources are clear, the timestamps are precise, and the results leave little room for interpretation.

This is likely the reason why Hyperliquid initially focused on the native market of regulated encryption based on objective and easily verifiable data sources. This approach reduces the risk of disputes early on and allows the infrastructure to be tested in simpler markets.

The permissionless phase will face significantly greater challenges. Deployers will need to define their own markets, select data sources, write settlement conditions, and handle corner cases. The greater the market's reliance on subjective or difficult-to-quantify events, the greater the risk of disputes.

In the outcome market, trust is not a secondary aspect, it is the product itself. A settlement error can cause reputational contagion that extends far beyond the market directly involved.

(2) Regulatory risks

Prediction markets remain a highly sensitive regulatory category. Depending on the jurisdiction, they may be considered financial products, gambling instruments, event contracts or derivatives. Polymarket, Kalshi, Crypto.com, Robinhood and Coinbase have all faced varying degrees of regulatory pressure on these issues.

For Hyperliquid, the risk profile will depend heavily on the types of markets allowed on HIP‑4. As long as the market remains crypto-native, objective, and tied to financial data, regulatory risks appear more manageable. If the ecosystem extends to politics, sports, or highly sensitive macro events, the situation becomes much more complicated.

Permissionless structures add another layer of complexity. Hyperliquid may only provide the infrastructure, but deployers and builders will likely play key roles in curation, distribution and compliance, especially as different jurisdictions require it.

(3) Information asymmetry

As a result markets naturally create financial incentives for actors with privileged information. This is a structural tension inherent in the model itself. Its strength lies in aggregating dispersed information, but it is also this feature that makes it attractive to participants with non-dispersed information.

This problem already exists in current prediction markets. Participants with faster access to data, insider information, or greater analytical capabilities may capture disproportionate profits. This is not always illegal because markets are fundamentally designed to reward informational advantage. However, once this information becomes privileged, confidential or improperly obtained, the nature of the risk changes completely.

On-chain transparency may help detect certain behaviors after the fact, but it cannot prevent them from happening. As HIP‑4 expands into more sensitive markets, monitoring mechanisms, the quality of market rules and the accountability of settlers will become key components of the system.

(4) Long tail quality issues

Permissionless deployment is an advantage, but it can also create noise. Poorly designed, illiquid, redundant, or improperly handled settlement markets can reduce the overall quality of the ecosystem. This risk is particularly acute for outcome markets—an ill-defined market not only creates a poor user experience but can also lead to controversy.

The economic viability of a deployer will therefore depend to a large extent on its control of market quality, not just market quantity. A good outcome market must be interesting, highly liquid, clearly defined, easy to settle, and cyclical enough to justify the attention invested by traders and market makers.

This is where Hyperliquid's structure may create an advantage. Slot recycling and the integration of HyperCore will naturally incentivize deployers to create a cyclical, economically practical market, rather than an infinitely extending "long tail of problems" with low liquidity. However, this incentive alone does not completely eliminate the risk of junk marketing or poor curation.

6. Conclusion

HIP‑4 is more than just Hyperliquid’s answer to Polymarket. It is a new primitive added to HyperCore and is completely consistent with the trajectory that the protocol has followed since its inception.

HIP‑1 introduces native assets. HIP‑2 enhances liquidity. HIP‑3 enables the creation of a permissionless perpetual contract market. HIP‑4 now opens the door to markets for discrete, conditional and non-linear returns. Each step follows the same logic: expand the range of markets tradable on HyperCore while maintaining the same execution engine, liquidity layer, and distribution infrastructure.

For prediction markets, HIP‑4 is likely to mark the transition we described in our previous article: from information markets to speculative markets and then gradually to risk markets. At this stage, markets no longer exist purely for prediction or betting, they become financial instruments designed to open positions, hedge or transfer event-driven risk exposure.

This is the fundamental difference between Hyperliquid and Polymarket or Kalshi. The latter two are still primarily consumer-facing distribution businesses built around attention, curation and user acquisition. Hyperliquid is building the financial infrastructure to enable composability of these markets with the rest of on-chain transactions.

The real question, therefore, is not whether Hyperliquid can capture some of the trading volume of existing prediction markets, but whether HIP‑4 can create an entirely new category of markets: smaller in number, better designed, more liquid, and ultimately of extremely high practical value to traders who seriously manage risk.