The asset accumulation scale of CEX is 100 times that of DEX, and the transaction volume is 10 times that of DEX. How does Hyperliquid ensure that it is not led by Binance? Is it the decentralized structure? Is it the algorithm?
What do we mean when we say “decentralized” exchanges? Which parts need to be “decentralized”? Assets? Matching? Prices? Clearing?
Introduction: A contract storm reveals the debate between centralization and decentralization
In March 2025, the JELLYJELLY contract on Hyperliquid caused a market turmoil.
In just a few hours, the price of the contract soared 429%, which is about to trigger a large-scale liquidation. If liquidated, the short position will be put into the on-chain liquidity vault HLP, causing a floating loss of tens of millions of dollars. The on-chain position is already in jeopardy, and the zone is in an uproar. At the same time, Binance is rare to launch JELLYJELLY perpetual contract trading "overnight"... It seems that it is about to be jointly beaten by CEX...
Finally, just before the nuclear bomb was about to be detonated, Hyperliquid validators urgently voted to intervene, forcing the delisting, liquidation, and freezing of transactions. For a time, everyone had a big question mark about the "decentralized" exchange?
This incident not only became a focus of endless debate in the crypto community, but also exposed a core problem:
On decentralized trading platforms, what determines prices? Who bears the risks? Are algorithms really neutral?
This article will use the JELLYJELLY incident as a guide to analyze the algorithm differences of the three major platforms in the core mechanisms of perpetual contracts - index price, mark price, and funding rate , and deeply analyze the financial philosophy and risk transmission mechanism behind them. You will see how different algorithms shape different trading styles, how they serve different types of traders, and how to determine whether you can escape the storm unscathed.
This is not only a technical dissection of a contract, but also a philosophical contest of market order design .
1. Overview of Perpetual Contract Trading
Before getting into the topic, let’s first clarify the three major components of perpetual contract trading:
Index Price : Tracks the price changes in the spot market and is the “theoretical anchor”. Hyperliquid calls it Oracle price.
Mark Price : The decisive price used to calculate key events such as unrealized profits and losses and liquidation.
Funding Rate : An economic mechanism that connects the spot and contract worlds, guiding contract prices to return to the spot world.
The following is an overview of the three platforms’ index price, mark price and funding rate algorithms:
If you really can't read so many words, just remember:
Whoever controls the Mark price has the power of life and death over the contract, so the core of Hyperliquid's "decentralization" lies in: how to ensure that the Mark price is not manipulated and can be verified.
Hyperliquid has made some optimizations based on the Binance algorithm, allowing prices to quickly return to market prices in extreme market environments and when Hyperliquid's on-site transactions are manipulated.
Hyperliquid has really put a lot of effort into avoiding outliers (pins).
2. The devil is in the details
The following is the dividing line between no (hard) and no (core) chatter, please bring your own kettle to prevent choking. Otherwise, it is recommended to jump directly to the next chapter.
A. Index Price / Oracle Price Comparison
The index price here in Hyperliquid is called the Oracle price. It is completely independent of its own market and is built by the validator node. It uses the weighted median method to combat extreme price fluctuations, which makes it more resistant to manipulation (plugging), but the update frequency is slow (once every 3 seconds). In layman's terms, it is to eliminate outliers and fluctuations to make the price smoother. The "slow" update frequency mentioned here is also a smoothing mechanism, which is not necessarily a bad thing. By the way, because of the characteristics of time difference, many smart scientists are also living on Hyperliquid.
B. Differences in Mark Price Mechanism and Algorithm Details
Binance 's mark price algorithm is based on the two principles of "price smoothness" and "market depth reflection". Its formula is based on the median of three types of prices: the bid/ask middle price of the contract market, the transaction price, and the impact price (Impact Bid/Ask Price). The impact price reflects the true cost of liquidity by simulating the impact of large market orders on the order book, avoiding being misled by the "pseudo-price" of the shallow market. Combined with the median construction after EMA processing, Binance's mark price changes smoothly and is resistant to plug-ins, which is suitable for large funds to stabilize the layout and institutional arbitrage strategies.
OKX takes a more "radical" approach, using only the bid/ask middle price in the market as the source of the mark price. This algorithm does not refer to the transaction price, nor does it consider the depth of the order book, so the price is extremely sensitive to small transactions and can easily cause drastic fluctuations due to large orders eating through the market. Although the volatility is higher, the price returns to the spot faster, so it is more suitable for high-frequency traders, pin-scalpers and short-term operations.
Hyperliquid 's mark price structure is a combination of Binance and OKX. It can be said to have "decentralized characteristics" in terms of data sources because it is controlled by several nodes. The nodes will combine three price sources to calculate the mark price:
150-second exponential moving average (EMA) of the difference between the Oracle price and the contract mid-price;
The median of the bid price, ask price and last traded price on the Hyperliquid platform itself;
The weighted median of the perpetual mid-price of multiple CEXs (Binance, OKX, Bybit, Mexc, Gate, etc.) (weights 3, 2, 2, 1, 1).
If any two of them fail, the system will supplement the median value processed by a 30-second EMA as a substitute.
The responsibilities of the on-chain validator are not only to update the Oracle and Mark Price regularly, but also to verify the integrity, timestamp, and deviation tolerance of the input source. This mechanism forms a certain degree of "algorithmic democracy" on Hyperliquid. Even the platform and validators cannot force the modification of the mark price logic, which greatly enhances the anti-manipulation ability.
Given that the mark price is the “only criterion” for determining whether a position is profitable/liquidated/exploded, “protecting” the mark price becomes the most important thing.
The core of Hyperliquid's "decentralization" is to tell everyone: I can't change the Mark Price to snipe your position, instead of on-chain matching, decentralized accounts, etc....
C. Funding rate algorithm and market behavior feedback mechanism
As the key economic lever that connects the spot and contract markets for perpetual contracts, the funding rate directly affects the degree of deviation between the contract price and the spot price, and guides the market to self-correct. The three major platforms have completely different technical paths and trading philosophies in the design of the funding rate algorithm.
In terms of the algorithm of funding rate, Hyperliquid introduced the Premium index (very similar to Bybit and Bitget) to the basic model of Binance's impact price (order book depth) and loan interest rate, and replaced the Oracle price in the calculation to be closer to the real market situation (Bybit and Bitget both use index prices). After all, under the same depth conditions, the meaning of the order book depth of 1u is different from the order book depth that rises to 3u or falls to 0.5u.
The premium index will be sampled every 5 seconds and calculated according to the hourly average (sampling 720 times per hour) to prevent short-term sharp fluctuations (spikes). The lending rate is fixed at 0.01%.
What are the problems or limitations of this algorithm? Yes, when the contract price deviates from the spot price, it will take a long time to "return". (Similar to Binance)
In order to make up for this shortcoming (Hyper does not have the huge amount of capital accumulation and arbitrage of Binance, and the price may not return in a short time), here are 3 settings with Hyperliquid characteristics
The first is the terrifying funding rate. In extreme cases, the funding rate may be 4%/hour, which is 96%/day. Moreover, the frequency and upper limit of the funding rate do not change with the assets.
The second is that the funding fee is based on the oracle price rather than the mark price, which is to prevent price manipulation to a certain extent. In an extremely strong situation where even if you manipulate the on-site price and there are no arbitrageurs to level the price, you can rely on the gradual return of the funding rate to the price.
The third is the frequency of charging funding rates
From the algorithm, Hyperliquid's funding rate is calculated every 8 hours, but when it is collected, it is indeed charged at ⅛ per hour. For example, assuming that the funding rate for 8 hours is 0.1%, then the long position will pay 0.0125% of the funding fee for the short position in the next hour. This "small steps and fast running" approach accelerates the "speed" of price return to make up for the limitations of the Binance-style order book depth model.
In contrast, Binance's funding rate relies on a longer settlement cycle (usually 8 hours), which is calculated by combining the order book depth to simulate the impact of large market orders on the buying and selling prices, while taking into account the borrowing rate (fixed at 0.01%). It is designed to provide institutional investors and medium- and long-term traders with a smoother and more predictable funding cost.
OKX's funding rate algorithm is relatively simple, based on the deviation of the bid and ask prices, and the settlement cycle is also relatively long. The lack of comprehensive consideration of the order book depth and borrowing costs leads to sharp fluctuations in funding rates, which is suitable for high-frequency, short-term aggressive strategies, but also brings a higher risk of liquidation.
Hyperliquid has found its own balance between Binance's "market structure" and OKX's "market orientation" - based on a relatively flat (order book depth and Premium) algorithm, coupled with high-frequency funding rate settlement and over-the-counter price feeds, supplemented by high funding fees in extreme situations, so that the on-site price can quickly return to the market price in extreme environments.
The algorithmic differences among these three have a profound impact on actual transactions.
3. Algorithms determine fate: trading strategies adapted to different platforms and the financial philosophy behind them
If the perpetual contract market is a silent war, then the price algorithms, liquidation logic and funding mechanisms of different platforms are the "doctrine" in this war. Three platforms, three philosophies, three types of traders, reflecting different financial literacy and values between the market mechanism and human nature. Let me reiterate that models and algorithms are not right or wrong, good or bad, but only about gains and losses .
Binance: Design of Institutional Rationalists (Quantitative Finance and the Efficient Market Hypothesis)
Binance's overall design tends to be "institutionalized and moderate", and its core concept is to "make the market predictable". This is highly consistent with the spirit of the quantitative finance school and the efficient market hypothesis (EMH), which assumes that the market is generally rational and can be tamed through statistical modeling.
Mechanism manifestation:
Mark price smoothing mechanism: Through the median, moving average and impact price concepts, the mark price is ensured to be stable and resistant to manipulation, reducing the liquidation risk caused by instantaneous fluctuations. This design provides a predictable liquidation threshold for large funds, enabling more robust position management.
Sophisticated modeling of funding rates: Combined with order book depth simulation of the impact cost of large market orders and fixed borrowing rates, its funding rate design is intended to provide a smooth and predictable funding cost. The longer settlement cycle (usually 8 hours) further reduces the impact of high-frequency funding cost fluctuations on medium and long-term strategies.
Risk buffer mechanism: Binance has a huge insurance fund to cover the losses of bankrupt positions, and activates the automatic liquidation (ADL) mechanism when the insurance fund is insufficient. These mechanisms are layered and have preset buffers for every node that may cause a systemic market collapse, aiming to minimize socialized losses.
Market behavior feedback: This robust modeling mindset attracts institutional investors and medium- and long-term traders who seek stable returns and controllable risks. They rely on the clarity and predictability of market rules to achieve steady growth of capital in a relatively stable environment through sophisticated quantitative models and arbitrage strategies. Binance's system is built for "institutional rationalists" who believe that the market can be managed and predicted through rigorous mathematical models and institutional design.
OKX: Design for Trading Instincts (Behavioral Finance)
OKX's strategy design is close to "fast, ruthless and accurate", and its philosophy is "the market is a reflection of human nature". This coincides with the logic of behavioral finance, which accepts that the market is irrational and traders are emotional. The truly smart people do not build smooth models, but break through those "assumptions" and find room for gaming in violent fluctuations.
Mechanism manifestation:
Sensitivity of mark price: If the mark price mainly depends on the bid/ask middle price, the price is extremely sensitive to small transactions and can easily fluctuate violently due to large orders. This design accelerates price discovery, but also increases the uncertainty of liquidation.
Volatility of funding rate: The funding rate algorithm is relatively simple and lacks comprehensive consideration of the order book depth, resulting in sharp fluctuations in its funding rate. Although OKX has tried to shorten the settlement cycle of some contracts (such as from 8 hours to 4 hours or 2 hours) to increase the responsiveness of the funding rate, this has further amplified its volatility.
Liquidation Mechanism: Liquidation logic is direct and fast, designed to quickly process risk positions.
Market behavior feedback: This mechanism naturally attracts high-frequency traders (HFT), "pin-splitting parties" and short-term traders. They are not afraid of volatility, but need volatility - because only "price disorder" is their source of profit. They are good at capturing instantaneous price deviations and taking advantage of insufficient market depth for rapid arbitrage or speculation. OKX's system is built for "trading instincts" who believe that the market is full of irrationality and are willing to fight in chaos.
Hyperliquid: On-chain structuralist design (institutional structuralism and code as law)
Hyperliquid is trying to create a new financial paradigm: decentralized governance + programmable price mechanism. Its philosophy is: algorithms are not about predicting the market, but about setting order. Rather than saying that Hyperliquid is an exchange, it is more like a financial protocol running on the blockchain - the price is determined by the consensus of the validator nodes, the position liquidation is supported by the HLP Vault, all transaction data is publicly available on the chain, and the platform is not allowed to operate in a black box, nor is the exchange allowed to "interfere" in the price and position processing.
I would like to add that the so-called privacy protection comes at a price, which is nothing more than a choice between your position and identity. If you choose to hide your identity, then your asset balance and position must be public and transparent, and vice versa. Whether it is centralized or decentralized, there is no good or bad, only whether it is suitable or not.
Mechanism manifestation:
Validator consensus price: The index price and mark price are aggregated from multiple CEXs by decentralized validator nodes, and methods such as weighted median are used to enhance anti-manipulation. The calculation logic of the mark price is ensured to be unchangeable by the on-chain consensus, which enhances the anti-manipulation ability.
HLP Vault: As a market-making and liquidation mechanism at the protocol level, Hyperliquid Liquidity Vault (HLP) aims to democratize institutional-level strategies and provide liquidity support for liquidation. This is a model in which the community shares risks and benefits.
High-frequency funding rates and extreme caps: The funding rate algorithm combines order book depth and external oracle prices, and accelerates market price regression by charging 1/8 of the 8-hour funding rate per hour, and a funding rate cap of up to 4%/hour in extreme cases. This design is designed to force rapid correction through the algorithm.
On-chain transparency: All transaction data and liquidation processes are publicly recorded on the chain, aiming to eliminate the black box operations of centralized exchanges.
Market behavior feedback: Hyperliquid practices an institutional structuralist financial philosophy - even if the world is chaotic, as long as the algorithmic rules are transparent and unchangeable, a relatively "neutral" market order can be constructed. It attracts traders who seek to rebuild a trust system through verifiable code and distributed governance. However, the JELLYJELLY incident reveals the tension of this philosophy in practice: although the logic of marking prices is designed to be tamper-proof, when the system faces a survival crisis, higher-level governance (validator node voting) can still intervene and override protocol behavior, which shows that "code is law" still needs to face the test of "rule by man" in extreme cases.
In summary:
Binance : Designed for "institutional rationalists", with clear rules and suitable for stable capital growth.
OKX : Designed for "trading instincts", fast and fierce, suitable for gamblers and traders
Hyperliquid : Designed for "on-chain structuralists", algorithms are paramount, suitable for on-chain arbitrage and extreme traders with large funds.
Conclusion: The end of the algorithm is the human heart
Price is the appearance of transactions; algorithm is the order of transactions. But whether it is Binance's institutional buffer, OKX's market behavior supremacy, or Hyperliquid's on-chain consensus, they are essentially trying to answer the same question: How can we trust an invisible market?
Some systems choose to use "stability" as an anchor, telling you that the rules will never change; some systems choose "volatility" as an anchor, believing that traders can always adapt to risks; and some systems try to write everything into the on-chain contract, no longer relying on people, but on transparent codes, distributed nodes, and cold formulas.
However, when the market is in an extreme situation, algorithms will exit and people must take the stage. The crisis of JELLYJELLY tells us that even the most decentralized system needs a temporary "central bank" to take over positions; even the most neutral consensus cannot completely eliminate disputes over "justice". Ultimately, prices are not determined by "algorithms", but by who we choose to believe .
Maybe we can't design a perfect system, but we can design a system that can constantly correct itself in imperfection. Binance uses institutions to buffer uncertainty, OKX uses gaming to stimulate market vitality, and Hyperliquid tries to establish a new consensus through governance and transparency. There is no right or wrong, only choice. After all, there is no perfect solution in this world.
Maybe we can't design a perfect system, but we can design a system that can constantly correct itself in imperfection. Binance uses institutions to buffer uncertainty, OKX uses gaming to stimulate market vitality, and Hyperliquid tries to establish a new consensus through governance and transparency. There is no right or wrong, only choice. After all, there is no perfect solution in this world.
In the future financial world, algorithms will continue to expand their territory. But we must realize that behind every logic written into the code is a shadow of value judgment.
You want freedom, but when your account is liquidated you want fairness;
You want fairness, but you want transparency when it comes to liquidation;
You want transparency, but you want to control everything when your account is liquidated.
Ultimately what you are chasing is not price, but the illusion of order.
People must always be responsible for their own values.
Let us always have a heart that respects the market.
Be like a Fujianese
