Oracle Management
Oracle infrastructure represents one of the most critical dependencies in DeFi lending protocols. FiRM's reliance on accurate, timely price feeds for collateral valuation and liquidation execution makes oracle performance a top priority for the RWG. Rather than treating oracle providers as external black boxes, the RWG maintains active partnerships to ensure feed configurations align with FiRM's specific risk requirements and operational needs.
Chainlink Partnership
The RWG works closely with Chainlink Labs to optimize price feed performance across all FiRM markets. This partnership extends beyond basic integration to include proactive configuration tuning, methodology evaluations, and post-incident improvements. Recent examples are shared below.
The RWG participates in oracle methodology assessments, particularly for stablecoin feeds where pricing approaches significantly impact manipulation resistance.
Volume-Weighted Average Price (VWAP) vs. State Price:
Traditional VWAP methodology aggregates trade volume across exchanges and data providers, but proved vulnerable to manipulation in low-liquidity stablecoin markets. Two notable incidents highlighted these risks:
deUSD on Avalanche: Large trade in low-liquidity pool caused sharp price spike captured by VWAP feeds
USD0 on Ethereum: 12-minute pool imbalance created $1.06 price spike representing majority market volume
Following RWG feedback and consultation with multiple protocols, Chainlink migrated deUSD and USD0 feeds to state price methodology in late 2024. State pricing uses liquidity depth in pools rather than trade volume, making it significantly more resistant to manipulation through large trades in thin markets.
Key State Price Features:
Liquidity-weighted rather than volume-weighted pricing
Lookback windows prevent sudden liquidity injections from skewing weights
End-of-block sampling prevents intra-block manipulation attempts
Multiple DEX aggregation reduces single-pool concentration risk
The RWG confirmed that USR/USD, sUSDe/USD, and crvUSD/USD feeds all utilize state price methodology, providing confidence in their manipulation resistance for high-leverage market deployment.
Deviation thresholds determine the minimum price movement required to trigger an oracle update. The RWG actively monitors these parameters across all consumed feeds and requests adjustments when market conditions or FiRM's risk profile warrant tighter responsiveness.
Successful Threshold Reductions:
USR/USD: Reduced from 1.0% to 0.5% (July 2025) to support high-leverage stablecoin markets with minimal margin for error
deUSD/USD: Reduced from 2.0% to 0.5% (July 2025) following partnership discussions and market growth
Additional Reviews: Ongoing evaluation of cbBTC/USD (currently 2.0%), CRV/USD, crvUSD/USD, and sUSDe/USD feeds for potential optimization
For FiRM's high-collateral-factor stablecoin markets, where users often maintain positions at 90%+ utilization through leverage looping, even 50 basis point price movements can trigger liquidations. Tight deviation thresholds are essential for protocol safety in these configurations.
Following the October 10th, 2025 market stress event, Chainlink together with the RWG identified opportunities to improve oracle responsiveness through Offchain Reporting parameter adjustments. OCR timing parameters (deltaRound, deltaGrace, deltaStage) control how quickly oracle nodes initiate new consensus rounds after detecting price changes.
October 2025 CVX Feed Reconfiguration: Post-event analysis revealed that the CVX/USD feed's 2% deviation threshold, combined with default OCR timings, contributed to update latency during the sharp drawdown. Rather than simply reducing the deviation threshold, Chainlink's data team recommended OCR reconfiguration to improve tail-latency performance - the critical moments when feeds must be most responsive.
The OCR upgrade enhances:
Faster reaction to volatility: Shortened deltaRound parameters enable the network to start new aggregation rounds sooner, capturing transient price moves that might otherwise be missed during rapid market swings
Improved tail performance: Testing demonstrated lower latency in worst-case scenarios (the critical 1% of update events), meaning prices update closer to real-time benchmarks even under network congestion
Protocol-level responsiveness: Unlike deviation threshold changes that simply trigger more updates, OCR tuning fundamentally improves how quickly the oracle detects and processes market shifts
This configuration approach addresses the root cause of update delays rather than treating symptoms, and has been extended to other high-priority FiRM markets.
Ongoing Monitoring & Communication
The RWG maintains regular communication channels with Chainlink's product, data, and engineering teams to ensure continuous optimization:
Feed monitoring: An alerts channel tracks new price feed deployments and deprecations
Risk-based prioritization: High-leverage and correlation-sensitive markets receive heightened scrutiny and faster parameter adjustment requests
Post-incident analysis: Collaborative forensics following market stress events inform configuration improvements
Proactive reviews: Periodic assessment of all consumed feeds against industry benchmarks and peer protocol configurations
This partnership model treats oracle infrastructure as a continuously optimized system rather than a static dependency, ensuring FiRM maintains best-in-class price feed performance as markets evolve.
Future Oracle Diversification
While Chainlink serves as FiRM's primary oracle provider, the RWG maintains exploratory discussions with alternative providers including Pyth Network, Chronicle, and RedStone. These conversations focus on:
Redundancy and fallback mechanisms for critical markets
Specialized feed types (pull oracles, high-frequency updates)
Emerging methodology innovations
Cost-efficiency for niche or experimental collateral types
Oracle diversification remains a long-term strategic consideration, balanced against the operational complexity and security implications of multi-oracle architectures.
Last updated