Gyro Consolidated Price Feeds

Introduction to Oracle System Models in DeFi

In decentralized finance (DeFi), oracle systems play a crucial role in providing external data to on-chain applications. However, the accuracy and reliability of these systems vary significantly, and each model comes with its own challenges. The primary models used are:

• Centralized oracles,

• Medianizing of several centralized nodes,

• DEX TWAPs, and

• Betting markets.

Understanding their strengths and weaknesses is essential for improving the security and efficiency of DeFi protocols.

1. Centralized Oracles : Centralized oracles rely on a single data provider, introducing a significant counterparty risk. The trust in a central entity means that if the provider's data is wrong, the system goes offline, or the provider behaves dishonestly, the oracle's data becomes compromised. This model requires trust in the provider's integrity and availability, making it susceptible to manipulation if the rewards outweigh the risks.

1. Medianizing of Several Centralized Nodes: It aggregates data from multiple oracle nodes and computes a median value. While this model reduces the risk from any single erroneous source, it still depends on the collective reliability of the nodes. The Medianizer only ensures the median of the input values is correct, not the actual data accuracy. In some variations, nodes can be penalized for deviating from the median, facing similar incentive issues as seen in betting market designs.

1. DEX TWAPs (Time-Weighted Average Prices): It uses historical data from decentralized exchanges (DEXs) to calculate average prices. Although this method provides on-chain observable prices and quantifiable manipulation costs, It struggles with limited liquidity and slower reaction to market changes. Prices are represented in terms of on-chain assets, which may not always align with real-world currency values, often requiring proxies like stablecoins. This dependency can shift, rather than solve, the oracle problem.

2. Betting markets: They rely on Schelling point games where participants vote on data feeds and face penalties for deviating from the consensus. This method can be problematic due to game-theoretic attacks, where participants might vote based on expected consensus rather than actual data. This creates a potential "beauty pageant" problem, where the consensus does not necessarily reflect the truth but rather the perceived majority opinion.