Oracle Risk

Oracles play a vital role on Tren Finance, as they do on most DeFi protocols, by providing the “true” price of an asset at a given time. For example, if asset A were to trade at $100 on one exchange while also trading at $98 on another exchange, an oracle will provide the “true” value of the asset based on its configuration. This price feed is used to trigger important functions like liquidations by providing data on what a user’s collateral is worth.

Due to the important role that oracles play, they are often the subject of price manipulation attacks. For example, if an asset does not have strong buy-side liquidity, an attacker could artificially inflate the price of an asset. The attacker could then use the asset to borrow an abnormal amount of trenUSD, leaving the protocol in debt. If an asset does not have strong sell-side liquidity, an attacker could artificially deflate the price of an asset, triggering liquidations. The attacker could then buy these liquidated positions at a discount. To mitigate against such attacks, a high level of importance is put on liquidity and both buy-side / sell-side price impact when evaluating assets, as shown in the risk methodolgy table. Another risk mitigation method is adjusting the maximum borrowable amount of trenUSD allocated for an isolated module. This figure will be set so that it does not exceed the potential cost of a potential price feed manipulation attack, and therefore should not be profitable.

As Tren Finance plans on listing a wide variety of assets and asset types, an oracle-agnostic approach is most appropriate for the protocol. Tren Finance will use a variety of oracles for price feeds, including Pyth, Chainlink, UniV3 (UniV4 upon release), and will continue to explore oracle solutions based on the different solutions that are required for each asset and isolated module. The different characteristics of each oracle, such as whether a TWAP or VWAP is used, will be assessed to analyze which oracle solution is most approapriate for a given situation.