Comparison of Contract Algorithms between CEX and DEX: Hyperliquid, Binance and OKX

In March 2025, the JELLYJELLY contract triggered a market turmoil on a certain decentralized trading platform. The contract price skyrocketed by 429% in a short period, almost triggering a massive liquidation. If liquidation occurred, short positions would be forced into the on-chain liquidity vault, resulting in huge floating losses. Meanwhile, a certain centralized exchange unusually and swiftly launched perpetual contract trading for JELLYJELLY.

As the crisis was about to erupt, the validators of the decentralized platform urgently voted to intervene, forcibly delisting, liquidating, and freezing transactions. This incident triggered doubts within the crypto community regarding "decentralized" exchanges and revealed a core issue: what are the determining factors for prices on decentralized platforms? Who bears the risk? Is the Algorithm really neutral?

This article will use this event as a starting point to compare and analyze the algorithmic differences in the core mechanisms of perpetual contracts among the three major platforms, exploring the financial philosophies and risk transmission mechanisms behind them.

Overview of Perpetual Contract Trading

Perpetual contract trading is mainly composed of three elements:

1. Index Price: Tracks the changes in spot market prices as a theoretical benchmark.
2. Mark Price: The decisive price used to calculate unrealized profits and losses, liquidation, and other key events.
3. Funding Rate: An economic mechanism that connects the spot and futures markets, guiding the futures price to revert to the spot.

The mechanism for controlling the mark price is the core of contract trading. The key to a certain decentralized platform lies in ensuring that the mark price is not manipulated and is verifiable. This platform has been optimized based on the algorithms of a certain centralized exchange, allowing prices to quickly revert to market prices in extreme market conditions.

Index Price/Oracle Price Comparison

The index price of a certain decentralized platform is referred to as the oracle price, completely independent of its own market and constructed by validator nodes. It uses a weighted median method to resist extreme price fluctuations, making it more resistant to manipulation, but the update frequency is slower (once every 3 seconds). This mechanism is designed to eliminate outliers and fluctuations, resulting in a smoother price.

Differences in Mark Price Mechanism and Algorithm Details

The marking price algorithm of a certain centralized exchange is based on two main principles: "price smoothness" and "market depth reflection". It combines the bid/ask mid-price, transaction price, and impact price of the contract market. This design allows for stable changes in the marking price, resistant to spikes, making it suitable for large capital stable layouts and institutional arbitrage strategies.

Another centralized exchange uses a relatively simple method, taking only the midpoint price of the bid/ask as the source of the marking price. This Algorithm is extremely sensitive to small trades, which can easily lead to sharp fluctuations, making it suitable for high-frequency traders and short-term operations.

The marking price structure of a certain decentralized platform integrates the above two methods. It is controlled by multiple nodes and combines three price sources: the exponential moving average of the oracle price and the price difference in the contract, the median of the platform's own best bid and ask prices along with the last transaction price, and the weighted median of perpetual prices from multiple centralized exchanges. This mechanism creates a certain degree of "Algorithm democracy," enhancing resistance to manipulation.

Funding Rate Algorithm and Market Behavior Feedback Mechanism

A certain decentralized platform has introduced a premium index in its funding rate Algorithm and uses oracle prices instead of index prices. To compensate for the shortcomings of price regression, the platform has implemented three unique settings:

1. In extreme cases, a funding rate of up to 4% per hour.
2. The funding fee is based on the oracle price rather than the mark price.
3. An eighth of the 8-hour funding rate is charged every hour.

This design aims to accelerate price reversion and address the limitations of order book depth patterns.

In contrast, the funding rate of a centralized exchange relies on a longer settlement period, combining order book depth and lending rates to provide institutional investors and medium to long-term traders with smoother funding costs.

The funding rate algorithm of another centralized exchange is relatively simple, calculated based on the deviation between the best bid and ask prices, leading to significant fluctuations in the funding rate, suitable for high-frequency and short-term aggressive strategies.

Trading Strategies Adapted to Different Platforms and the Financial Philosophy Behind Them

1. A centralized exchange: The design of rational system builders

   • Core Concept: "Make the market predictable"
   • Mechanism: smooth marking price, refined funding rate modeling, risk buffer mechanism
   • Suitable for institutional investors and medium to long-term traders who seek stable returns and controllable risks.

2. Another centralized exchange: The design of trading instinct.

   • Core concept: "The market is a reflection of human nature"
   • Mechanism: Sensitive mark price, highly volatile funding rate, rapid liquidation
   • Suitable for: high-frequency traders, "pinning party" and short-term traders

3. A decentralized platform: On-chain structuralist design

   • Core Concept: Algorithm Sets Order
   • Mechanism: Validator consensus price, on-chain liquidity treasury support, high-frequency funding rate, on-chain transparency
   • Suitable for: Traders seeking to rebuild trust systems through verifiable code and decentralized governance.

Conclusion

Different trading platforms attempt to answer how to establish trust in the market through their own algorithms and mechanisms. Whether pursuing stability, embracing volatility, or relying on on-chain consensus, each approach reflects specific value judgments and market philosophies. In extreme cases, human intervention may still be necessary, revealing the limitations of purely algorithmic systems.

The future financial world will continue to be shaped by algorithms, but we must recognize that every algorithm contains value judgments. Traders need to choose platforms that suit their own risk preferences and trading strategies while maintaining a sense of respect for the market.