Introduction

In Web3, building a decentralized derivatives trading platform is no longer a novel concept. From early pioneers like GMX, dYdX, and Synthetix to the recent rise of Jupiter, Drift, and Hyperliquid, numerous protocols have continuously attempted to capture this emerging market, leading to increasingly fierce competition. However, due to limitations in blockchain technology, security, and user experience, decentralized derivatives platforms, while capable of attracting on-chain traders, still struggle to compete with centralized exchanges as the preferred choice for most users. Despite these challenges, Hyperliquid has emerged as a rising star. Within just over a year since its launch, it has gained significant traction by offering a fast and seamless trading experience, capturing more than half of the market share in a short time. Its highest single-day trading volume exceeded $11 billion, earning it the reputation of being the “on-chain Binance” within the community, positioning itself as a formidable challenger to centralized exchanges. This article will take a technical perspective to explore how Hyperliquid has built a dedicated Layer 1 blockchain to enhance transaction speed and user experience, compare it with other decentralized derivatives trading Layer 1 solutions like dYdX, and summarize the key factors behind Hyperliquid’s success along with the challenges it must address moving forward.

Hyperliquid Introduction

Hyperliquid is a Layer 1 blockchain based on the HyperBFT consensus mechanism designed to build a high-performance on-chain financial system. Its primary product is a decentralized derivatives exchange that utilizes a Central Limit Order Book (CLOB) system. Order placement, matching, and settlement all occur on-chain, ensuring transparency in transaction records. With the ability to process 100,000 orders per second and a block confirmation time of less than one second, it delivers a trading experience comparable to that of centralized exchanges.

In terms of liquidity, Hyperliquid introduces the Hyperliquidity Provider (HLP), which adopts an active market-making strategy as a counterparty to trades, allowing regular users to contribute funds to HLP for market-making and earn profits. Additionally, Hyperliquid’s CLOB system also supports spot trading, where besides listing its native token, other projects can compete for listing slots through a Dutch auction mechanism.

With high performance, low latency, and transparent on-chain records, Hyperliquid enables traders to execute a diverse range of quantitative trading strategies while maintaining the decentralized advantages of Web3. Currently, Hyperliquid accounts for over 70% of the total trading volume among decentralized derivatives platforms. The team is also actively developing HyperEVM, aiming to integrate more DeFi protocols from the EVM ecosystem and further expand Hyperliquid’s ecosystem.

Trading volume share of each decentralized derivatives trading platform (Source: Artemis)

Team and Financial Background

Hyperliquid was founded by Harvard alumni Jeff and iliensinc, with team members hailing from institutions such as Caltech and MIT, and possessing work experience at renowned companies like Airtable, Citadel, Hudson River Trading, and Nuro. The team brings extensive expertise in quantitative trading, blockchain technology, and user experience design. They began operating a crypto market-making business in 2020 and expanded into the DeFi sector in the summer of 2022. Recognizing the market inefficiencies, technical shortcomings, and poor user experience in many DeFi projects, they set out to develop a high-performance Layer 1 blockchain that genuinely meets user needs. Notably, Hyperliquid has not raised external funding; all operational capital has been self-funded by the team, allowing them to remain independent from external capital influences and stay fully focused on building a platform that truly serves its users.

Hyperliquid technological strengths

Hyperliquid’s technological strengths lie in its high-performance Layer 1 blockchain and efficient decentralized derivatives trading platform. The synergy between these two aspects contributes to its success. This section breaks down Hyperliquid into these two components, examining the standout features of each.

Layer 1

As a Layer 1 blockchain, its consensus algorithm determines its performance ceiling, a critical factor for DeFi protocols. Inspired by HotStuff, Hyperliquid developed its own HyperBFT, optimized for end-to-end latency. For clients in close geographic proximity, the median latency from order placement to confirmation is only 0.2 seconds, with the 99th percentile latency at 0.9 seconds. This low-latency performance enables Hyperliquid to support high-frequency trading and complex financial operations. Although the official documentation does not disclose detailed technical aspects of HyperBFT, insights can be drawn from HotStuff, a variant of Byzantine Fault Tolerance (BFT) consensus, which improves on traditional BFT by addressing two key issues:

• Reducing Communication Complexity

Traditional BFT algorithms experience exponential growth in communication complexity as the number of nodes increases. HotStuff optimizes this by aggregating multiple signatures into a single signature, significantly reducing communication overhead and storage requirements. This alleviates network load and enhances scalability.

• Pipeline Design

Block proposal processes are divided into three stages—prepare, pre-commit, and commit—allowing the network to handle multiple blocks at different stages simultaneously. New proposals do not need to wait for previous ones to complete all stages before processing begins. This minimizes node idle time, keeping them continuously operational and significantly improving network efficiency, transaction speed, and block generation.

Thanks to these improvements, HotStuff enhances blockchain performance, throughput, and scalability, making it particularly suited for applications requiring rapid finality. It has been adopted by notable blockchain projects such as Meta’s former Web3 initiative Libra and its successor, Aptos. HyperBFT, as a derivative of HotStuff, inherits these advantages, positioning Hyperliquid as a high-performance blockchain. Over the past three months (from November 17, 2024, to February 17, 2025), Hyperliquid has processed over 33 billion transactions, achieving an average TPS of 4,180.

Hyperliquid trading volume (Source: Hyper Stats)

Currently, the network has only 25 validator nodes, with the top five largest nodes operated by Hyper Foundation, collectively accounting for nearly 80% of the total staked amount.

Hyperliquid validator (Source: Hyperliquid)

HyperEVM

As early as May last year, Hyperliquid announced the development of HyperEVM to achieve Ethereum-compatible smart contract functionality, and on February 18 of this year, it officially launched HyperEVM. According to the official documentation, HyperEVM is not an independent blockchain but is built on Hyperliquid Layer 1, benefiting from HyperBFT consensus protection. This integration allows it to interact with native Layer 1 components such as spot and perpetual contract order books, enabling seamless asset circulation.

As illustrated, Hyperliquid DEX operates on a Rust-based virtual machine, providing spot and derivatives trading services, while HyperEVM runs parallel to the Rust VM, enabling Ethereum-compatible smart contracts with the following characteristics:

• Permissionless
  Unlike Rust VM, which is not fully open to the public—currently limiting dApp development and requiring a Dutch auction for token issuance—HyperEVM adopts an open architecture. Developers can freely create applications and issue both fungible and non-fungible tokens on HyperEVM without prior approval, fostering innovation and expanding the ecosystem.

HyperEVM frame diagram (Source: ASXN)

• Interoperability
  HyNative tokens created under HIP-1 on Hyperliquid can circulate between Rust VM and HyperEVM, and HyperEVM can utilize the Rust VM’s oracle services. However, assets on HyperEVM cannot be fully interpreted with Rust VM unless they have obtained minting rights through a Dutch auction on Rust VM. This imposes certain limitations on interoperability.

• Gas Fee
  Currently, smart contracts and spot trading on Rust VM do not require users to pay gas fees. However, on HyperEVM, transactions require $HYPE as gas fees. According to Hypurr’s latest statistics, over 140 protocols have already been deployed on HyperEVM.

Currently, according to Hypurr’s statistics, more than 140 protocols have been deployed on HyperEVM.

HyperEVM blockchain (Source: Hypurr)

Hyperliquid DEX

Contract Trading

Unlike most DEXs that use AMM models, Hyperliquid adopts a Central Limit Order Book (CLOB) system similar to CEXs, ensuring that all order placements, executions, and settlements are recorded on-chain. While CLOB allows for more precise trading strategies, its multiple transaction steps can lead to network congestion if each action requires on-chain validation. However, as previously mentioned, Hyperliquid’s HyperBFT consensus mechanism significantly enhances network performance, making it capable of handling the transaction volume associated with CLOB.

For liquidity provision, Hyperliquid introduces the Hyperliquidity Provider (HLP) vault, which functions as a market maker within the order book system, supplying liquidity and executing liquidations while accumulating platform fees. The HLP’s trading positions, order book activities, transaction history, and earnings are fully transparent. Users can deposit funds into the HLP to participate in market-making and earn proportional returns based on their capital contribution. Additionally, users can create or join alternative liquidity vaults with different market-making strategies to optimize profits.

HLP details and information (Source: Hyperliquid)

Hyperliquid vault (Source: Hyperliquid）

Spot Trading

Although contract trading is Hyperliquid’s primary business, it also offers spot trading, distinguishing itself from other derivatives exchanges while maintaining the same CLOB system for order execution. The HIP-1 standard governs Hyperliquid’s native token issuance. Any project wishing to list a token on Hyperliquid must acquire deployment rights through a public Dutch auction. The auction process operates as follows:

The auction runs every 31 hours, and anyone can participate. The initial bidding price starts at twice the previous auction’s final price and then decreases linearly to 10,000 USDC. The first bidder secures the right to deploy a token. The figure below shows the auction prices of each token as of February 20.

Hyperliquid auction prices of spot tokens (Source: ASXN Data)

Additionally, for certain spot tokens with relatively low liquidity, Hyperliquid has introduced HIP-2 to provide initial liquidity support. HIP-2 is similar to the CLMM mechanism adopted by many DEXs, allowing liquidity provision within a specified price range. However, unlike CLMM, HIP utilizes an order book system, where liquidity-providing orders are automatically adjusted every 3 seconds with each block update, maintaining an order range of 0.3% at each level. HIP-2 consists of five parameters:

• the token name (spot)
• the initial price for order placement (starPx)
• the number of orders (nOrders) which are set as sell orders by default
• the size of each order (orderSz), and the seeded levels (nSeededLevels), which convert a portion of the sell orders closest to the initial price into buy orders. For example, if the number of orders is set to 10 and the seeded levels to 2, then the two orders closest to the initial price will be buy orders, while the remaining eight will be sell orders. More detailed explanations are provided in the example below.

Suppose there is a token $TEST with starPx = 1 (USDC), nOrders = 5, orderSz = 2, and nSeededLevels = 2. Given that the price gap between each adjacent order is fixed at 3%, the initial order distribution provided by HIP-2 is as follows:

• buy orders at 1 and 1.003 (since nSeededLevels = 2, the two closest orders to the initial price are buy orders)
• sell orders at 1.006, 1.009, and 1.012.

In this scenario, the market price of $TEST at launch is 1.0045. As trading commences, the orders provided by HIP-2 are gradually consumed by other traders. Every 3 seconds, HIP-2 replenishes new buy and sell orders at the same price levels based on the status of existing orders and available funds to ensure sufficient liquidity.

It is important to note that the seeded level (nSeededLevels) plays a crucial role in maintaining liquidity. In the above example, with the seeded level set at 2, even if $TEST faces immediate selling pressure upon launch, HIP-2 can still provide buy orders at 1 and 1.003 to prevent a rapid price crash. However, the USDC required for buy orders must be provided by the token deployer and will be permanently locked in HIP-2. Conversely, if the deployer fails to set an appropriate seeded level according to the token’s distribution, the token price may fall beyond HIP-2’s liquidity range. Furthermore, HIP-2 only provides the most basic liquidity support, and other market participants can place orders within the system, allowing both mechanisms to coexist without conflict.

Comparison Hyperliquid and dYdX

Hyperliquid and dYdX are both decentralized derivatives exchanges built on Layer 1 blockchains, each holding a significant share of the market. Therefore, the following comparison will analyze the similarities and differences between Hyperliquid and dYdX from two key perspectives: the underlying operational structure of their Layer 1 foundations and their derivatives trading mechanisms.

Operation Structure

Hyperliquid’s ability to implement an on-chain CLOB system is attributed to the high performance enabled by its HyperBFT consensus algorithm, allowing the network to process a large volume of transaction requests. In contrast, dYdX, another derivatives-focused Layer 1 blockchain, is built on the Cosmos SDK framework and utilizes the CometBFT algorithm. Compared to HyperBFT, which is based on HotStuff, CometBFT has a higher communication complexity and does not employ a pipeline-based workflow, resulting in slightly lower throughput and slower block generation.

Although dYdX’s consensus algorithm optimization is not as advanced as Hyperliquid’s, it enhances network performance by adopting an “off-chain order matching, on-chain execution” approach. Specifically, each dYdX validator node stores the order book locally off-chain. When a user places an order, one of the validator nodes broadcasts the information to other nodes, updating their respective order books in memory. The transaction is recorded on-chain only when an order match occurs, at which point it undergoes validation and settlement. This means that users do not need to pay gas fees when submitting or canceling orders on dYdX. The order execution process follows these steps:

1. The user places an order through the website or another front-end interface.
2. The order is transmitted to a validator node, which then broadcasts it to other nodes, updating their local order books.
3. Orders are matched, and the matched order is included in a proposed new block.
4. The new block undergoes the consensus process. If more than two-thirds of the nodes vote to approve the block, it is stored in all validators’ on-chain databases; otherwise, the block is rejected, and a new one is proposed.
5. Once the block is confirmed and recorded, the data flows from the nodes to the indexers, which update the front-end interface with the latest information.

dYdX Chain Operation Structure (Source: dYdX Docs)

As a result, orders on dYdX are only recorded on-chain when they are successfully matched, which effectively reduces network load and increases the number of transactions that can be processed. According to the official documentation, dYdX’s system is capable of processing approximately 500 orders per second, which is about 100 times higher than the typical transactions per second. This capacity is expected to continue growing in the future.

Derivative Trading

Similar to Hyperliquid, dYdX also utilizes an order book system. However, unlike Hyperliquid, not every trading action takes place on-chain—only order matching or liquidation executions are recorded on the blockchain. In terms of liquidity, dYdX has implemented MegaVault, which functions similarly to Hyperliquid’s HLP in providing liquidity across various markets. Users simply need to deposit USDC, and MegaVault will automatically allocate these funds to different sub-markets, matching them with corresponding orders. The profits generated are then distributed proportionally among all participants. Notably, unlike Hyperliquid, where the liquidity pool is managed by the official team, MegaVault’s operator is determined through community voting. The current market maker is Greave Cayman Limited.

MegaVault (Source: dYdX)

Unfortunately, aside from MegaVault, dYdX does not offer users the ability to create their own vaults, as Hyperliquid does. This limitation restricts the variety of market-making strategies available, making dYdX less flexible in terms of user choice.

Summary

Overall, Hyperliquid’s adoption of the HyperBFT consensus mechanism enables an on-chain CLOB system, ensuring that all trading activities are recorded on-chain while delivering an exceptional trading experience. This approach is significantly more efficient than dYdX’s CometBFT. Although dYdX attempts to compensate for its performance limitations through “off-chain matching and execution,” it still lags behind Hyperliquid in terms of throughput and speed. Additionally, in terms of liquidity, Hyperliquid provides users with a diverse range of market-making options, whereas dYdX only offers a single MegaVault, with its market maker determined by community voting. Considering both aspects, Hyperliquid demonstrates clear advantages in performance and operational flexibility.

However, dYdX exhibits a higher degree of decentralization compared to Hyperliquid. Currently, the development of the dYdX protocol is entirely governed by community voting, with the official foundation not acting as the primary decision-maker. Additionally, all protocol revenues are allocated to the community treasury, ensuring a high level of transparency. In contrast, Hyperliquid has only opened HyperEVM to developers, while its codebase and technical details remain undisclosed. Furthermore, the largest staking nodes are operated by the Hyperliquid Foundation, raising concerns about excessive centralization.

Potential Risks

Although Hyperliquid has achieved significant success in the market within a short period, there are two major risks worth noting:

• Over-centralization
  The five largest validator nodes in the Hyperliquid network are all operated by the Hyperliquid Foundation, accounting for 80% of the total staked assets. This means that the network’s operations are almost entirely controlled by the official entity. Such a high concentration of power introduces multiple risks, including potential validator misconduct in transaction ordering, governance decisions being dominated by the foundation, and community voting becoming a mere formality. For a Web3 ecosystem that emphasizes decentralization and transparency, Hyperliquid’s high degree of centralization raises significant uncertainties about the network’s future development.

• Capital Risk
  Currently, Hyperliquid only accepts USDC deposits from Arbitrum, with all user funds stored in a bridge contract connecting Hyperliquid and Arbitrum. This means that if the contract is compromised by hackers, all assets on Hyperliquid could be stolen. When users request withdrawals, the contract requires approval from at least two-thirds of the signing authorities. However, there are only four validators with signing privileges, meaning that if a hacker gains control of three or more private keys, they could execute withdrawals at will. Although approved withdrawals enter a dispute period of approximately 200 seconds—during which suspicious transactions can trigger a system lock on the bridge contract to halt withdrawals—the restriction can ultimately be lifted through a vote by the validator set. With Hyperliquid’s Total Value Locked (TVL) exceeding $2.5 billion, having such a large amount of assets stored in a single bridge contract controlled by a small number of validators poses a considerable financial risk.

• Code Risk
  The underlying Layer 1 code of Hyperliquid is not open-source. While this reduces the likelihood of external attacks, it also prevents the broader community from reviewing the code for security vulnerabilities. As transaction volumes continue to grow and HyperEVM becomes operational, the potential for undiscovered code vulnerabilities will only increase as Hyperliquid’s ecosystem expands.

Conclusion

Overall, Hyperliquid’s success can be attributed not only to its outstanding product performance and user experience but also to its strong emphasis on community engagement. Without any venture funding, Hyperliquid allocated 31% of its total token supply for a genesis airdrop, which helped attract substantial capital and users after the token launch. This strategy also fostered developer participation on HyperEVM, positioning Hyperliquid as the most popular decentralized derivatives exchange in the market. Looking ahead, while Hyperliquid must continue working towards greater decentralization, the growth and development of the HyperEVM ecosystem will also play a crucial role in determining its long-term influence. Its future progress remains a key area of interest.