Felix
Felix will be launching feUSD, a CDP stablecoin on HyperEVM once mainnet goes live, and additional money markets soon after. Users will be able to borrow against collateralized assets without intermediaries, unlocking more financial workflows (e.g capture the funding rate, run a carry trade, etc.) and cross-margin capabilities. At the same time, these systems must confront an array of risks: from collateral price volatility and liquidity gaps to manipulative exploits and more. We touched on some of these tangential ideas in our previous piece focused on protocol security. Now we turn our attention to risk management.
Broadly speaking, overcollateralized money market protocols remain solvent and functional by ensuring every user position’s collateral value remains higher than its outstanding borrow value. When any position becomes undercollateralized—i.e., when the loan-to-value (LTV) drifts dangerously close to or exceeds safe thresholds—liquidations occur. Liquidation participants and bots close unhealthy positions (or the user self-manages), thereby protecting the entire protocol from cascading defaults and a buildup of bad debt.
Fundamentally, the core goal of any money market is to minimize bad debt, which means ideally having no user position where the borrowed value or debt value is greater than the collateral value. Of course, there are complexities to ensuring that this happens. Each collateral type carries a unique risk profile. Supply caps, borrow caps, interest rates, and liquidation incentives are all adjustable parameters that create a delicate interplay between maximizing user adoption (growth) and protecting the protocol from losses (value at risk). In this article, we explore our risk management strategy—particularly in the context of Hyperliquid’s native oracle, orderbook ecosystem, and other quirks / nuances specific to Felix’s position as Hyperliquid-native from Day 1.
The job of money market risk management is to continually balance growth and VaR (value at risk). The Felix risk team must monitor changing market conditions, adjust parameters, and swiftly plug leaks that arise.
We define below:
Value at Risk (VaR): This is the protocol’s potential losses, typically from bad debt incurred if borrowers default and the collateral cannot sufficiently cover the borrowed amount. If a money market takes on aggressive parameters, the market will benefit with more liquidity, capital efficiency, and other positive business metrics, but could take on additional protocol risk.
Growth: On the other side of the coin, the protocol thrives on capital efficiency, i.e., how effectively it can put user deposits to productive use. Higher TVL (total value locked) and robust borrow/supply ratios reflect a thriving market, but come with additional exposures if volatility strikes.
Balancing these two competing forces is a constant job. Our view is that DeFi risk management blends three key elements:
Quantitative Analysis: Historical volatility, liquidity metrics, and market simulations form the bedrock of numerical assessments.
Qualitative Insights: Sometimes, signals of imminent risk emerge from governance changes, macro shifts, or unexpected exploits. Knowledgeable contributors who “sense” trouble can be just as important as raw data.
Monitoring & Vigilance: DeFi operates 24/7 in a permissionless manner. Constant risk reviews, parameter updates, and war-room readiness are essential to staying ahead of emergent threats.
Below, we share some Hyperliquid-specific factors and challenges that the risk team must navigate.
2.1 Oracle Reliability
Hyperliquid features a native price oracle, but its codebase remains relatively fresh and subject to iterative updates. Just recently, the Hyperliquid team pushed an update where EVM apps can fetch prices via the pre-compiles. This constant change in the testnet tells us that the codebase is yet to be stable and production-ready. Indeed, production-readiness is critical and not a trivial guarantee. In the “dark forest” of DeFi, a single oracle glitch can mean catastrophic protocol losses, especially around thinly traded assets.
2.2 Spot Orderbook Liquidations
Unlike some EVM-native approaches that rely on automated market makers (AMMs) for on-chain liquidations, Hyperliquid also enables spot orderbook liquidations with deep liquidity from market makers. While deeper liquidity can reduce slippage for partial closeouts, it’s not always atomic—especially if a pre-compiles EVM-based liquidation function doesn’t exist at launch. The lack of atomicity introduces potential risk if prices move unfavorably in between the liquidation transaction and the final settlement.
2.3 Liquidator Preparedness
A strong network of independent, well-incentivized liquidators forms the backbone of any lending protocol’s solvency. For Felix on Hyperliquid, the plan is to have:
An In-House Liquidator Bot: Developed by Felix to ensure a baseline level of liquidation coverage.
A Partnership with Manhattan Research: A proprietary trading firm that will run liquidation services on day one.
Multiple Community-Run Bots: Decentralizing the liquidation process, reducing single points of failure.
2.4 New, Tested Assets: HYPE and LST HYPE
Newly introduced assets like HYPE and its liquid-staked version (LST HYPE) introduce further considerations. These tokens are akin to L1 tokens (like ETH or SOL) with additional staking mechanics. While yield and capital efficiency are appealing, the protocol must account for:
Reduced Liquidity: Staked variants can have even thinner liquidity than HYPE, which can complicate liquidations in the early days.
Oracle Nuances: If the LST’s price depends on a redemption rate or a combination of staked/unstaked markets, oracles become more complex.
Slashing Risks: Slashing events can abruptly cut token value, which increases price volatility.
Because these assets don’t have an extensive track record, the risk team needs to remain extra vigilant with feUSD mint caps against these assets and provide conservative LTVs until we have more data.
2.5 Cross-Margin Functionality
Hyperliquid’s robust user base across spot and perpetual trading sets the stage for advanced cross-margin workflows. Many will likely turn to Felix or similar protocols for overcollateralized lending, bridging capital between multiple trading venues. While synergy and capital efficiency can flourish, complex interdependencies (a hallmark of DeFi “lego” composability) can introduce systemic risk that isn’t well-researched from the Ethereum ecosystem or Solana ecosystem. Monitoring positions that have exposure to several markets simultaneously becomes crucial—especially under stress scenarios.
Money market protocols rely on various parameters to ensure system health. Each parameter requires extensive work.
3.1 Loan-to-Value (LTV) Ratios
When deciding collateral-specific LTVs, risk managers weigh factors such as:
Collateral Volatility: Large price swings increase the likelihood a position collapses if the price drops too quickly. Monitoring trailing 30-day and 90-day volatility helps predict stability.
Liquidity Depth: If an asset can be liquidated quickly on major DEXs or specialized orderbooks, the protocol can tolerate slightly higher LTVs because liquidators can exit more easily without steep slippage.
Market Health: Global market sentiment, volumes, and velocity of price changes over time (e.g., trailing 30-day or 90-day snapshots) can tilt these risk parameters in a more conservative or aggressive direction.
3.2 Mint Caps
Felix will implement mint caps on feUSD at launch to ensure that Felix scales in a risk-adjusted way. Felix mint caps will be used to ensure that there is no excess feUSD minted against relatively illiquid assets as well as secondarily to ensure that a significant amount of feUSD can always be swapped for USDC without excess slippage in the early days.
3.3 Liquidation Incentives
Along with LTVs and mint caps, the Felix risk team must help determine liquidation incentives on day one to ensure that liquidations are always profitable to avoid a buildup of bad debt. Liquidation incentives must factor in the volatility of the approved collateral assets as well as their liquidity in a similar way to the way by which LTVs are assessed.
To deliver high quality risk management across feUSD and other future Felix money markets, we are working with the Anthias Labs risk arm – they will serve as the specialized party focusing on Felix risk needs.
4.1 The Risk Audit
Before any official rollout or major parameter shift, Anthias conducts a risk audit that dives into core questions:
Exotic Collateral Onboarding: How do we safely onboard HYPE LSTs or bridged assets? For bridged majors, what is the bridging mechanism’s trust model?
Oracle Reliability: Are we confident in the Hyperliquid native oracle? Is there any documented downtime or missed price updates?
Network Performance: Could network congestion (akin to Ethereum gas spikes) hamper timely liquidations?
External Variables: We also look at the broader DeFi ecosystem for regulatory changes, cross-chain bridging vulnerabilities, or macro events that could destabilize collateral valuations.
4.2 Collateral Onboarding Checkpoints
When greenlighting a new collateral asset, risk teams like Anthias will follow a structured checklist:
Secondary Liquidity for Liquidations: Is there enough daily trading volume and order book depth?
Risk Vectors: For tokens bridged from another chain, can that bridge be compromised? For stablecoins, is the custodian or algorithmic mechanism battle-tested?
Parameter Selection: Based on the data, set cautious mint caps and LTV thresholds.
Audit and Oracle Maturity: Verified oracles and at least two recognized smart contract audits.
4.3 Ongoing Monitoring & Due Diligence
Similar to protocol security, our risk management approach doesn’t end after launch day. It extends into continued monitoring systems that will be built out by our team in collaboration with Anthias Labs to track the following:
Systemic Risk: Track how users might loop multiple assets or cross-collateralize positions in ways that create hidden correlation.
Market Metrics: Observe big transactions, watch for abnormal liquidation events, and regularly check the health of the feUSD peg (if relevant).
Qualitative Signals: Keep tabs on various product developments across asset issuers, L1 developments, etc.
4.4 Simulations and Stress Testing
In addition to ongoing monitoring and due diligence, simulation and stress testing are needed. For examples of these models that will be employed on day one:
Agent-Based Simulations: Model how lenders, borrowers, and liquidators behave under different parameter settings or market events.
Monte Carlo & GARCH Models: Generate realistic price trajectories to gauge tail risk, letting the protocol see how it might fare in Black Swan scenarios.
Scenario Stress Tests: Replay historical events—like a May 2021 or March 2020 crash—to see if the protocol can weather the storm without incurring widespread bad debt.
4.5 Final Risk Recommendations
Following these tests, the risk team compiles continual reports, offering specific actions such as:
Recommended LTV Adjustments
feUSD mint Cap Changes
Onboarding/Offboarding Certain Collateral Assets
Reserves and Liquidation Incentive Updates
This iterative process—data analysis, stress testing, parameter tuning—will serve as the base of Felix’s risk strategy.
To do risk management well, our view at Felix is that “set it and forget it” is a recipe for disaster. Instead, protocols like Felix must emphasize a multilayered, ongoing approach. This entails:
Expertise & Culture: A dedicated risk team that truly cares—showing up daily, scanning for red flags, and swiftly responding to anomalies.
Robust Infrastructure: Sound oracles, multiple liquidation bots, stable bridging solutions.
Adaptive Parameters: Willingness to pivot from high growth to defense parameterization (and vice versa) based on real-time conditions.
It’s our job to constantly iterate upon our process and methodology – combining quantitative inputs, qualitative judgment, and continual monitoring to serve as the foundation for Felix’s risk platform. This is because Felix is building for the long-term and must be able to weather the inevitable volatility of DeFi markets—particularly as we look to pioneer new territory on Hyperliquid.
