The Battle of Contract Algorithms Between CEX and DEX: Hyperliquid, Binance, OKX

In March 2025, the JELLYJELLY contract triggered a market turmoil on a certain decentralized trading platform. Within just a few hours, the contract price skyrocketed by 429%, almost triggering large-scale liquidations. If liquidations occurred, short positions would be forced into the on-chain liquidity vault, resulting in floating losses in the tens of millions of dollars. As on-chain positions were on the brink of collapse, a certain centralized trading platform quickly launched perpetual contract trading for JELLYJELLY, further escalating the situation.

On the eve of a crisis about to erupt, validators of decentralized platforms urgently voted to intervene, forcing the delisting, liquidation, and freezing of transactions. This event sparked doubts within the crypto community about "decentralized" exchanges, while also exposing a core issue: on decentralized trading platforms, how are prices determined? Who ultimately bears the risk? Is the Algorithm really neutral?

This article will take this event as a starting point to analyze the algorithmic differences in the core mechanisms of perpetual contracts among the three major platforms—index price, mark price, and funding rate, and to explore the financial concepts and risk transmission mechanisms behind them in depth. We will see how different algorithms shape trading styles, serve different types of operators, and how they determine traders' survival ability in market storms.

This is not only an analysis of contract technology but also a philosophical contest about market order design.

Overview of Perpetual Contract Trading

Perpetual contract trading consists of three main elements:

1. Index Price: Tracks changes in the spot market price and serves as a "theoretical anchor." Some platforms refer to it as oracle price.

2. Marked Price: The decisive price used to calculate unrealized profits and losses, liquidation, and other key events.

3. Funding Rate: An economic mechanism connecting the spot and contract worlds, guiding contract prices to revert to the spot.

The algorithm differences of the three major platforms in these core mechanisms are as follows:

Index Price/Oracle Price Comparison

The index price of a certain decentralized platform is referred to as the oracle price, which is completely independent of its own market and constructed by validator nodes. It uses a weighted median method to resist extreme price fluctuations, exhibiting strong anti-manipulation properties, but the update frequency is slower (once every 3 seconds). This design aims to eliminate outliers and volatility, resulting in a smoother price. The slower update frequency also serves as a smoothing mechanism and is not necessarily a drawback.

Mark Price Mechanism Differences and Algorithm Details

The marked price algorithm of a certain centralized platform is based on the two principles of "price smoothness" and "market depth reflection". It uses the median of the bid/ask midpoint, transaction prices, and impact prices in the contract market. The impact price reflects the true liquidity cost by simulating the effect of large market orders on the order book. The median constructed after processing with EMA ensures that the marked price changes smoothly and is resistant to spikes, making it suitable for stable allocations of large funds and institutional arbitrage strategies.

Another centralized platform adopts a more aggressive approach, using only the bid-ask midpoint as the source of the mark price. This algorithm is extremely sensitive to small trades and is prone to severe fluctuations due to large orders penetrating the order book. Although the volatility is higher, the price reverts to the spot market faster, making it more suitable for high-frequency traders, pinning players, and short-term operations.

The mark price structure of the decentralized platform integrates the two methods mentioned above. It is controlled by multiple nodes, taking into account the EMA of the difference between the oracle price and the price in the contract, the median of the platform's own best bid and ask prices, and the perpetual average weighted median from multiple centralized exchanges. This mechanism forms a certain degree of "algorithm democracy," enhancing the resistance to manipulation.

Funding Rate Algorithm and Market Behavior Feedback Mechanism

The decentralized platform introduced a premium index in the funding rate Algorithm, sampling every 5 seconds and calculating the hourly average to prevent short-term drastic fluctuations. To compensate for the slow price return speed, the platform has adopted three distinctive settings:

1. In extreme cases, a high funding rate of up to 4% per hour can be reached.
2. The funding fee is based on the oracle price rather than the marked price to prevent price manipulation.
3. The funding rate is charged one eighth every hour, accelerating the price return.

The funding rate of a certain centralized platform relies on a longer settlement period (usually 8 hours), simulating the impact of large market orders on the buy and sell prices in conjunction with the order book depth, while also considering a fixed borrowing rate. This design provides institutional investors and medium to long-term traders with a smoother and more predictable cost of capital.

The funding rate algorithm of another centralized platform is relatively simple, calculated based on the deviation between the bid and ask prices on the order book, with a similarly long settlement cycle. The lack of comprehensive consideration of order book depth and borrowing costs leads to dramatic fluctuations in the funding rate, making it suitable for high-frequency, short-term aggressive strategies, but also bringing higher liquidation risks.

Algorithm Determines Fate: Trading Strategies Adapted to Different Platforms and the Financial Philosophy Behind Them

The pricing algorithms, clearing logic, and funding mechanisms of the three major platforms reflect different financial philosophies and values.

A certain centralized platform: The design of rational institutionalists

The overall design of the platform is inclined towards "institutionalization and moderation", with the core idea being "making the market predictable". This is highly compatible with the quantitative finance school and the efficient market hypothesis, assuming that the market is generally rational and can be tamed through statistical modeling.

Mechanism embodiment:

• The mark price smoothing mechanism ensures price stability and resistance to manipulation.
• The funding rate fine modeling provides smooth and predictable funding costs.
• With a large insurance fund and an automatic position reduction mechanism, it reduces social losses.

This design attracts institutional investors and medium to long-term traders who pursue stable returns and controllable risks.

Another centralized platform: The design of trading instinct.

The platform's strategy design approaches "fast, fierce, and accurate", and its philosophy is that "the market is a reflection of human nature". This aligns with the logic of behavioral finance, accepting the reality of market irrationality and traders' emotionality.

Mechanism embodiment:

• The mark price is extremely sensitive to small trades and can cause significant fluctuations.
• The funding rate fluctuates wildly, lacking a comprehensive consideration of the order book depth.
• The clearing logic is direct and swift.

This mechanism attracts high-frequency traders, "pinning groups", and short-term traders who excel at capturing instantaneous price deviations and seeking profits amidst chaos.

Decentralized platform: The design of on-chain structuralists

The platform attempts to create a new financial paradigm: decentralized governance + programmable pricing mechanism. Its philosophy is: the Algorithm does not predict the market, but rather establishes order.

Mechanism reflection:

• Validator consensus price enhances anti-manipulation capability
• The liquidity vault acts as a market-making and clearing mechanism at the protocol layer.
• High-frequency funding rates and extreme caps accelerate market price return
• All transaction data and settlement processes are publicly recorded on the blockchain.

This design appeals to traders seeking to rebuild trust systems through verifiable code and distributed governance. However, in extreme cases, the test of "human governance" still needs to be faced.

Conclusion

Different trading platforms attempt to address the core question of how to trust an invisible market through their own algorithm designs. Some systems anchor on stability, while others choose volatility, and some try to write everything into on-chain contracts. However, when the market is in extreme situations, algorithms may withdraw, and human factors still play an important role.

Ultimately, the price is determined not only by the Algorithm but also by who we choose to believe to make that determination. While it is impossible to design a perfect system, we can create a system that continuously self-corrects in imperfection. In the future financial world, the Algorithm will continue to expand its territory, but every piece of logic written into the code casts a shadow of value judgment.

Let us always maintain a sense of awe towards the market and recognize that people ultimately have to take responsibility for their values. In the pursuit of freedom, fairness, and transparency, we are actually seeking an illusion of order. Regardless of which trading platform one chooses, it is essential to carefully consider the underlying design philosophy and its impact on trading strategies.