Summary
This report outlines LlamaRisk’s activities for H2 of 2025 (July–December). As a service provider to Curve DAO, LlamaRisk conducts research and risk management across Curve products, with a continued focus on ecosystem safety, market optimization, and long-term protocol development. This report is intended to provide transparency into our work, methodology, and governance engagement over the period.
During this half-year, our efforts centered on proactive risk management during periods of market volatility, research and preparation for LlamaLend v2, and continued optimization of both crvUSD stability and LlamaLend market performance. In parallel, our work with asset issuers, improvements to oracles, and the introduction of a structured market deprecation framework strengthened Curve’s resilience and operational robustness as the protocol continues to scale.
Key Highlights:
- Grants: Secured a grant of 250,000 OP tokens from Optimism to be used as incentives for bootstrapping the LlamaLend v2 deployment
- Revenue Growth: Protocol Revenue has been increasing steadily
- LlamaLend v2 Preparation: Coordinated with Swiss Stake on audit preparation, launch sequencing, and technical specifications for the upcoming v2 deployment
- New Collateral Research: Developed risk frameworks and oracle solutions for novel collateral types, including Pendle Principal Tokens (PTs). And investigating Curve AMM LP positions that can be used as collateral in v2 to support yield-bearing and structured assets at launch.
- Crisis Management: Successfully navigated the October 10th flash crash and Elixir depeg events with minimal protocol impact, demonstrating robust risk management frameworks
- crvUSD Stability: Contributed towards improved peg stability through refined monetary policy parameters and PegKeeper management
- Market Optimization: Implemented market deprecation framework, successfully applied on Sonic chain, resulting in ~50% TVL reduction of underperforming markets
- Oracle Infrastructure: Upgraded Oracle implementations for fxSAVE and sreUSD LlamaLend markets, improving price feed robustness and security
- Governance Leadership: Submitted 26 proposals across LlamaLend, crvUSD, and AMM categories with strong community engagement (39 replies, 2,763 views)
The LlamaRisk team continues to secure the Curve ecosystem through rigorous research and proactive governance engagement. We remain grateful for the community’s continued support and trust.
Governance Activity
LlamaRisk engages with the Curve community through consistent, transparent, and well-documented governance participation. Each proposal is accompanied by clear reasoning and supported by both qualitative research and quantitative analysis. Our governance activity reflects the depth of our research and provides the rationale behind the recommendations we bring to the DAO.
LlamaLend-Related Proposals
LlamaLend v2 is approaching, and we have been actively researching frameworks to safely onboard a broader range of collateral types, including Pendle PTs. This work focuses on understanding asset-specific risk profiles and their implications for market design.
In parallel, LlamaRisk applies this research directly to existing and upcoming LlamaLend markets to improve safety and capital efficiency. Our governance efforts include identifying and remediating markets with unsafe or fragile oracle setups, calibrating market parameters to balance utilization and stability in the face of changing market conditions, and supporting the launch of new markets through informed gauge voting and Resupply proposals. These actions aim to ensure that market growth is aligned with robust risk controls and long-term protocol health.
Highlights:
- Improved market safety by updating liquidation parameters and oracle configurations across multiple LlamaLend markets on Mainnet, Arbitrum, Optimism, and Fraxtal.
- Stabilized utilization and borrow rates through targeted monetary policy and discount parameter optimizations.
- Introduced a standardized framework for safely deprecating low-activity LlamaLend markets, first applied on Sonic, resulting in a ~50% reduction in TVL.
- Strengthened oracle robustness in yield-bearing and structured asset markets by replacing or updating price feed implementations.
| Topic | Type | Category | Date |
|---|---|---|---|
| Deprecate All LlamaLend Markets on Sonic | Llamalend | Proposals | Dec 2025 |
| Modify Liquidation Discounts in LlamaLend Markets CRV-long (Mainnet) and ARB-long (Arbitrum) | Llamalend | Proposals | Nov 2025 |
| Kill Gauges on All Elixir Markets (deUSD & sdeUSD) | Llamalend & AMM | Gauge Proposals | Nov 2025 |
| Kill Gauges for LlamaLend iBTC-crvUSD Market and iBTC-WBTC Pool on Arbitrum | Llamalend & AMM | Gauge Proposals | Oct 2025 |
| Update Oracles for sreUSD LlamaLend Market | Llamalend | Proposals | Oct 2025 |
| Kill Gauge for the Old LlamaLend sDOLA Market on Mainnet | Llamalend | Gauge Proposals | Oct 2025 |
| Increase Loan Discount to 20%, LlamaLend CRV-long on Mainnet | Llamalend | Proposals | Oct 2025 |
| Modify Loan Discounts in LlamaLend Markets on Arbitrum, Optimism, and Fraxtal | Llamalend | Proposals | Oct 2025 |
| Set EMAMonetaryPolicy on fxSAVE LlamaLend Market on Mainnet | Llamalend | Proposals | Oct 2025 |
| Modify WBTC-long and wstETH-long LlamaLend Markets on Mainnet MonPol Parameters | Llamalend | Proposals | Sep 2025 |
| Adjust CRV-long LlamaLend Market Rates: Pt. 1 | Llamalend | Proposals | Sep 2025 |
| Modify Rates on Arbitrum WETH and WBTC LlamaLend Markets | Llamalend | Proposals | Sep 2025 |
| Adjust sDOLA-long LlamaLend MonPol Params | Llamalend | Proposals | Sep 2025 |
| Set EMAMonetaryPolicy on sUSDe-long-v2 LlamaLend Market on Mainnet | Llamalend | Proposals | Sep 2025 |
| Modify WETH2-long LlamaLend Market on Mainnet min/max Rates | Llamalend | Proposals | Sep 2025 |
| Add sreUSD LlamaLend Market to the Gauge Controller | Llamalend | Gauge Proposals | Sep 2025 |
| Add sdeUSD LlamaLend Market to the Gauge Controller | Llamalend | Gauge Proposals | Jul 2025 |
crvUSD and Peg Stability
During this period, LlamaRisk continued to focus on maintaining crvUSD stability through parameter calibration, PegKeeper management, and monetary policy research. This work has become increasingly important as crvUSD usage has expanded and flow dynamics have evolved, including the growing influence of external routing and integrations such as YieldBasis. These integrations introduce larger, more frequent interactions with PegKeepers, underscoring the importance of careful parameter tuning and ongoing monitoring to preserve peg resilience across varying market conditions.
Highlights:
- Improved crvUSD peg stability by adjusting TargetFraction and rate0 parameters to reduce rate volatility
- Strengthened peg defense by onboarding the frxUSD/crvUSD PegKeeper with a 3M debt ceiling and updating liquidation and loan discounts in the WETH-Mint market
- Conducted R&D on a new Monetary Policy that uses EMA smoothing on the debt fraction to further reduce rate volatility.
| Topic | Type | Category | Date |
|---|---|---|---|
| Adjust Parameters (TargetFraction and rate0) for crvUSD Monetary Policy | crvUSD | Proposals | Nov 2025 |
| Add frxUSD/crvUSD to PegKeepers with 3M Debt Ceiling | crvUSD | Proposals | Aug 2025 |
| Investigation of EMA Smoothing Debt Fraction on the crvUSD Monetary Policy | crvUSD | Proposals | Dec 2025 |
| Update Liquidation & Loan Discounts for WETH-Mint Market | crvUSD | Proposals | Dec 2025 |
AMM Related Proposals
AMM Proposal Summary
LlamaRisk continues to actively manage AMM pools, focusing on gauge hygiene, parameter optimization for strategic pairs, and supporting new asset integrations. This period emphasized removing gauges from deprecated or risky assets while optimizing pools that serve key protocol functions.
Highlights:
- Gauge Hygiene: Removed CRV emissions from deprecated pools
Governance Activity Table
| Topic | Type | Category | Date |
|---|---|---|---|
| Kill CRV Emissions to yETH/ETH Pool | AMM | Gauge Proposals | Dec 2025 |
| Kill Gauge Emissions to InceptionLRT Pools | AMM | Proposals | Jul 2025 |
| Kill Gauge to Polygon Pools That Use Aave V2 aTokens | AMM | Proposals | Jul 2025 |
| Kill Gauges on All Elixir Markets (deUSD & sdeUSD) | Llamalend & AMM | Gauge Proposals | Nov 2025 |
| Kill Gauges for LlamaLend iBTC-crvUSD Market and iBTC-WBTC Pool on Arbitrum | Llamalend & AMM | Gauge Proposals | Oct 2025 |
Key Metrics
Protocol Revenue has substantially increased, especially with the introduction of YieldBasis
crvUSD Peg has remained tightly pegged through 2025, especially compared to historical performance
Summary Statistics
- Total Proposals: 24 distinct topics
- Total Engagement: 39 replies, 2,826 total views
- Average Engagement: 1.6 replies per topic, 118 views per topic
- Most Engaged Topic: “Adjust Parameters (TargetFraction and rate0) for crvUSD Monetary Policy” (513 views, 2 replies)
- Proposal Types: Standard Proposals (18), Gauge Proposals (6)
Research and Methodology Updates
This period marked significant advancements in LlamaRisk’s research capabilities and methodological frameworks, particularly in preparation for LlamaLend v2 and the development of enhanced risk management tools.
Pendle PT Markets Research
In preparation for LlamaLend v2, we conducted comprehensive research into Pendle Principal Token (PT) markets as potential collateral assets. This research examined:
- Price Discovery Mechanisms: How PT prices converge to underlying assets at maturity and implications for oracle design
- Liquidity Characteristics: PT market depth, trading volumes, and liquidity concentration across different underlying assets
- Maturity Risk: Framework for assessing safe debt ceilings based on time-to-maturity and underlying asset volatility
- Oracle Design: Specialized oracle contract for time-decaying assets with converging price paths
- Risk Parameters: Appropriate liquidation discounts, loan-to-value ratios, and debt ceilings for PT collateral
- Technical Requirements: Advised the Swiss Stake team on FE and BE adjustments needed to handle PTs.
This research will enable Curve to capture market share in the rapidly growing markets while maintaining our risk standards.
LlamaRisk Debt Ceiling Methodology
We have developed a comprehensive debt ceiling methodology that will now be applied across Curve’s lending infrastructure. This methodology considers multiple risk factors, including:
- Collateral liquidity depth
- Borrower behavioral patterns
- Unique features of LlamaLend markets (LLAMMA liquidation mechanism)
- Global leverage across major venues such as Aave
Current Implementation:
- Being integrated for the upcoming LlamaLend v2 which introduces borrow caps at the contract level
Future Development:
We are actively automating the debt ceiling calculation process across all markets. This automation will enable us to:
- Rapidly respond to changing market conditions
- Provide more frequent parameter updates
- Reduce manual calculation overhead
- Improve consistency across all markets
- Enable proactive risk management
This automated system will be particularly critical for LlamaLend v2, where contract-level borrow caps require more dynamic management as market conditions evolve.
Market Deprecation Framework
We developed and successfully implemented a comprehensive framework for deprecating underperforming LlamaLend markets. This framework includes:
- Criteria for Deprecation: Analysis and research on parameter and monetary policy updates designed to avoid unfair treatment of existing users while being impactful enough to encourage gradual position unwinds and deter new users from entering.
- User Communication Protocol: Timeline and channels for notifying affected users
- Technical Implementation: Process for gauge removal, parameter changes to discourage new positions, smart contract deployments, monitoring during wind-down, and necessary FE & BE adjustments to ease the process
- Risk Management: Ensuring user safety during the migration period
- Monetary Policy Contract: A tailored monetary policy, a custom smart contract, and a safe way for deployment were developed by Llamarisk.
The Sonic market deprecation served as our proof-of-concept, demonstrating that markets can be safely wound down while protecting user interests.
Monetary Policy Research
We continued advancing our understanding of optimal monetary policies for different collateral types:
- EMA Smoothing Investigation: Researched applying exponential moving average smoothing to the debt fraction component of the crvUSD monetary policy to reduce rate volatility
- Yield-Bearing Asset Optimization : Implemented and refined EMAMonetaryPolicy for assets like fxSAVE, building on successful implementations in sfrxUSD and sDOLA markets
- Rate Responsiveness : Analyzed borrower behavior across different interest rate levels to identify thresholds at which borrowing activity meaningfully reacts, informing calibration of rate curves rather than optimizing short-term rate stability metrics.
Tooling and Infrastructure
Oracle Development & Infrastructure
LlamaRisk advanced oracle capabilities across both new collateral types for v2 and existing market improvements:
- PT Oracle: Developed a specialized oracle for Pendle Principal Tokens that handles time-decaying price convergence toward underlying asset value at maturity, enabling safe PT collateral onboarding at v2 launch
- LP Price Oracle: Contributed to the strategy for pricing LP positions as collateral, expanding v2’s supported asset types
- Structured Asset Oracles: Tailored implementations for assets like sreUSD and fxSAVE with complex pricing mechanisms, replacing previous implementations to reduce oracle volatility
- Safety Check Enhancements: Additional validation layers to prevent oracle manipulation
- Monitoring: Developed methodology and internal tooling for monitoring Oracle setups using Curve AMM as price source, enabling systematic health checks across markets
LLAMMA Simulator Enhancement
We have made significant progress on the LLAMMA Simulator, transforming it from a one-time analysis tool into a comprehensive continuous monitoring system in collaboration with Swiss Stake.
Recent Improvements:
- Code Refactoring: Improved code structure for better maintainability and extensibility
- Debugging Enhancements: Added comprehensive logging and error handling
- Performance Optimization: Reduced simulation time for faster parameter testing
- Scenario Coverage: Expanded range of market conditions and stress scenarios
- Methodology Research: Optimized the methodologies used in calculating the parameters
Continuous Monitoring System:
We are now working towards transitioning the simulator into a continuous monitoring system for:
- Periodic Parameter Calculation: Calculates optimal
loan_discountandliquidation_discountparameters for all markets - Market-Specific Analysis: Tailors parameters to each market’s unique characteristics (oracle, volatility, liquidity, collateral type)
- Alert System: Flags markets where current parameters may be suboptimal
This continuous monitoring approach allows us to be more proactive in parameter adjustment, ensuring that all LlamaLend markets maintain appropriate risk parameters as market conditions evolve. The system will be particularly valuable as we scale to support more markets and collateral types with LlamaLend v2.
IPOR Vault Integration
We have created two IPOR vaults that supply crvUSD to carefully curated markets:
- Mainnet Vault: Allocates to LlamaLend markets, StakeDAO’s vaults that allocate to LlamaLend and Savings crvUSD on Ethereum mainnet
- Arbitrum Vault: Allocates to LlamaLend markets through StakeDAO’s vaults on Arbitrum
Key Features:
These vaults utilize StakeDAO’s only-boost, which dynamically allocates capital between StakeDAO and Convex based on which platform yields higher returns. This intelligent allocation strategy serves multiple strategic purposes:
- Rate Stabilization: Helps stabilize interest rates across LlamaLend markets by providing a consistent supply
- Adoption: Boost crvUSD adoption by introducing a new supply sink.
- scrvUSD Rate Alignment: The mainnet vault allocates to scrvUSD and LlamaLend markets through intelligent capital allocation, which will have an increasingly stabilizing impact as TVL grows
- Migration Support: Designed to facilitate smooth capital migrations once LlamaLend v2 launches
- Market Liquidity: Provides reliable liquidity depth for borrowers across multiple markets
LlamaRisk charges no fees on these vaults, prioritizing ecosystem growth and market stability.
This infrastructure positions LlamaLend markets to maintain healthy liquidity and competitive rates as we transition to v2, while demonstrating the composability and integration potential of Curve’s lending platform.
YieldBasis Integration Impact
The integration of YieldBasis (YB) has materially changed the flow profile of the crvUSD system. From a revenue perspective, YieldBasis has been a clear positive: it has driven sustained swap volume through crvUSD pools and increased fee generation accruing to Curve DAO. From a stability perspective, however, its impact is more nuanced and state-dependent.
DEX fee revenue increased following the YieldBasis launch, underscoring the revenue-generating potential and scalability of crvUSD infrastructure when paired with external demand drivers.
LlamaRisk’s analysis shows that YieldBasis introduces large, frequent, and highly responsive flows through PegKeeper pools, making PegKeepers more actively engaged in day-to-day peg management. These flows are not uniformly peg-supportive. While YieldBasis often defends the peg by buying crvUSD below parity, it also frequently sells crvUSD during below-peg conditions, mechanically expanding supply at precisely the wrong time. The result is not persistent depegging, but higher short-term peg stress, expressed primarily through increased volatility in mint rates and more frequent PegKeeper inventory adjustments.
Importantly, empirical analysis indicates that during economically active periods, YieldBasis routing explains the dominant share of large PegKeeper reserve changes, with reserve movements and YB flows matching almost one-for-one. This implies that YieldBasis is not merely an incremental source of noise, but a first-order marginal driver of PegKeeper inventory dynamics when active. As a consequence, PegKeeper reserve depth—and not just aggregate liquidity—has become a binding constraint for short-term peg defense.
These observations directly motivated LlamaRisk’s work on crvUSD rate stability during the quarter. In particular, smoothing mechanisms and monetary policy adjustments were designed to absorb the increased frequency of PegKeeper interventions induced by YieldBasis activity, reduce feedback loops between reserve depletion and borrow rates, and preserve peg resilience during periods of elevated market volatility.
Overall, YieldBasis has strengthened crvUSD’s economic throughput while simultaneously increasing the importance of careful PegKeeper liquidity management and monitoring. A comprehensive analysis report detailing all of the observations and insights will be published soon.
Source: Tokenterminal
Source: Valueverse
Crisis Management: October Flash Crash
On October 10th, a significant market flash crash tested DeFi protocols’ resilience. Despite the extreme market conditions, Curve’s infrastructure demonstrated resilience with minimal overall impact. Following the event, LlamaRisk conducted a thorough impact assessment and implemented parameter adjustments to strengthen market resilience.
Actions Taken:
- Impact Evaluation: Assessed bad debt exposure across markets,
- Parameter Adjustments: Submitted proposals to make market parameters more conservative, including updates to liquidation and loan discounts across Arbitrum, Optimism, and Fraxtal markets
- Gauge Management: Killed gauges on affected Elixir markets (deUSD & sdeUSD) following depeg events
These post-event adjustments improved safety margins across affected markets and informed our ongoing parameter optimization.
Looking Ahead: Q1 2026 and Beyond
LlamaLend v2 Launch Preparation
Our primary focus for Q1 2026 is supporting the successful launch of LlamaLend v2:
- Launch Coordination: Working closely with Swiss Stake on deployment and preparation of Llamalend v2
- Grant: Activating OP grant to be used as incentives for bootstrapping Llamalend v2
- Pendle PT Integration: Finalizing parameters for PT markets based on completed research
Grants & Incentives
LlamaRisk plans to actively pursue additional grants and incentive programs aligned with Curve’s long-term roadmap. These grants will primarily support LlamaLend v2 adoption, risk tooling development, and monitoring infrastructure, helping fund work that directly improves protocol safety, scalability, and market resilience. Where appropriate, incentives will be structured to bootstrap new markets responsibly while maintaining Curve’s risk standards and avoiding short-term or misaligned growth.
Risk Management Enhancement
In 2026, LlamaRisk will further shift risk management from periodic review toward continuous monitoring and proactive intervention across Curve’s lending and stablecoin systems. As the number of markets and asset types expands with LlamaLend v2, maintaining safety requires more frequent, data-driven oversight rather than ad-hoc parameter updates.
Our focus areas include:
- Continuous Market Monitoring
Ongoing monitoring of LlamaLend and crvUSD markets - Automated Risk Signals
Development of quantitative thresholds that flag when markets move outside of expected or safe ranges, enabling earlier identification of emerging risks before they materialize into stress events. - Systematic Parameter Review
Using monitoring outputs to inform more regular and structured parameter updates, reducing reliance on reactive changes during periods of volatility. - Scenario-Driven Risk Assessment
Expanding the use of simulation and stress testing to continuously evaluate market behavior under adverse conditions, with results feeding directly into risk buffers and market-specific configurations.
This approach aims to improve protocol resilience by identifying risk earlier, standardizing response mechanisms, and ensuring that market growth remains aligned with Curve’s long-term safety and stability objectives.
Delegation Proxy & Adaptive Risk Management
As Curve’s lending infrastructure scales, reliance on governance votes for routine, low-risk parameter adjustments introduces latency that can meaningfully impact market safety during rapidly changing conditions. To address this, a Delegation Proxy model is being introduced, where clearly scoped control over specific market parameters can be delegated to a trusted risk manager under predefined rules.
Under this model, the delegated party would be responsible for:
- Making bounded parameter updates (e.g., loan discounts, liquidation discounts, rate curves)
- Acting based on observable market data and predefined risk thresholds
- Operating under full transparency and post-hoc accountability
LlamaRisk intends to take an active role in this framework, leveraging continuous monitoring, automated risk signals, and simulation-driven analysis to safely execute parameter updates when required. The monitoring and tooling work described above is a prerequisite for such a system, ensuring that any delegated actions are systematic, data-driven, and aligned with governance intent rather than ad-hoc judgment.
H2 2025 marked significant progress in preparing Curve’s lending infrastructure for its next phase of growth. The LlamaLend v2 launch in Q1 2026 will be a key milestone, supported by the research, tooling, and governance frameworks developed during this period.
We thank the Curve community for continued support and welcome feedback on this report. For integrations support or risk consultations, reach out via our public Telegram channel or contact @WormholeOracle on TG.