Oracle design and DeFi risk: lessons from the USDe depeg
In October 2025, USDe traded as low as $0.65 on Binance during what became the largest liquidation event in DeFi history. In lending protocols using fixed-price oracles, nothing happened. In protocols using market-based oracles, positions approached liquidation thresholds before the price recovered. Same asset, same collateral. Different outcomes based entirely on oracle configuration.
This is why oracle design is a risk factor, not a neutral technical setting.
What happened on Binance
The depeg was exchange-specific. USDe’s underlying fundamentals remained intact throughout the event. Ethena’s reserves, fully verifiable on-chain, were not impaired. The $0.65 price reflected a temporary dislocation on a single venue, driven by a concentrated mass of liquidations compressing Binance’s order book.
Market-based oracles that aggregate prices across multiple sources reported a very different number. RedStone’s USDe/USD price feed, which aggregates 14 data sources, reported a low of approximately $0.996 during the same period: a marginal discount to parity, not a 35% collapse. Fixed-price oracles, which hold USDe at a constant $1.00 regardless of market conditions, showed no movement at all.
The same event produced three different price signals depending on where you looked: $0.65 (Binance spot), $0.996 (aggregated market oracle), $1.00 (fixed-price oracle).
Two oracle designs, two risk profiles
Fixed-price and market-based oracles represent different risk trade-offs, not a hierarchy where one is categorically better.
Fixed-price oracles maintain a constant valuation for the collateral token. For assets with deep reserves, verifiable backing, and a track record of peg stability, this design reduces the risk of liquidations triggered by temporary market dislocations (events that resolve before they create actual capital loss). The October 2025 USDe event was this type: exchange-specific, short-lived, and fundamentally disconnected from reserve quality.
Market-based oracles aggregate live prices across multiple venues and produce a feed that moves with the market. For assets where a price decline might indicate genuine reserve deterioration rather than a temporary dislocation, a market-based oracle ensures liquidations happen before losses compound.
The problem is that these two scenarios look identical in the moment. A price declining on one exchange could be a temporary dislocation or early evidence of a fundamental failure. Oracle design determines which of these scenarios a protocol acts on, and which it ignores.
Why oracle design is part of the risk assessment
Oracle design is one of five independent dimensions in Credora’s risk framework. The rating for a lending market reflects both the collateral’s characteristics and how the oracle prices that collateral, not just one or the other.
A market with sound collateral and a mismatched oracle carries different risk than a market with the same collateral and an appropriate oracle. Fixed-price oracles are not inherently high risk. But they require specific conditions to justify their use: liquid reserves that can be verified independently, a transparent and auditable backing mechanism, and an asset with a stable peg across market cycles. When those conditions are met, a fixed-price oracle reduces unnecessary liquidation risk. When they are absent, the same design delays necessary ones.
Credora’s assessment remained unchanged through the October 2025 event because the methodology uses an indexed price aggregated across multiple sources. Isolated exchange incidents do not disproportionately influence the output. The methodology uses multi-source aggregation, filtering exchange-specific noise without ignoring genuine price moves. As described in What is a DeFi risk rating?, oracle risk is assessed independently from collateral quality, smart contract risk, and liquidity.
Key takeaway
Oracle design is not a neutral technical setting. Fixed-price and market-based oracles make different assumptions about the nature of a price move: temporary dislocation or fundamental deterioration. The October 2025 USDe depeg on Binance illustrated the difference: aggregated market-based oracles reported a price near $1.00 while Binance showed $0.65. Neither was wrong. They were optimized for different risk scenarios. For vault curators and protocol builders, the question is whether the oracle configuration matches the collateral’s risk profile. A risk rating assesses both together. See How Credora rates a DeFi vault for the methodology.
Frequently asked questions
What is the difference between a fixed-price oracle and a market-based oracle in DeFi?
A fixed-price oracle maintains a constant valuation for a collateral token regardless of market conditions. A market-based oracle aggregates live prices across multiple venues and produces a feed that tracks the market. Fixed-price oracles reduce liquidation risk from temporary dislocations; market-based oracles capture genuine deterioration faster. The right choice depends on the nature of the collateral and the verifiability of its reserves.
Did the USDe depeg on Binance trigger liquidations in DeFi lending protocols?
It depended on the oracle configuration. Protocols using fixed-price oracles saw no movement. Protocols using market-based oracles that aggregate across multiple sources saw prices near $0.996, which did not breach typical liquidation thresholds. The $0.65 Binance low was not reflected in aggregated oracle feeds because it was an isolated venue dislocation, not a reflection of USDe’s actual reserve value at the time.
How does oracle design affect a DeFi risk rating?
Oracle design is one of five independent dimensions in Credora’s risk framework. A lending market’s rating reflects both the collateral’s characteristics and how that collateral is priced by the oracle. Fixed-price oracles matched to the right collateral reduce unnecessary liquidation risk; mismatched configurations can delay necessary ones. The overall rating captures this interaction, not just the collateral quality in isolation.