The Problem
DeFi lending protocols separate risk measurement from risk enforcement. Parameters are calibrated ex ante through analysis and governance, then enforced at fixed thresholds until the next governance update. This architecture has fundamental limitations that become binding as assets move along the pricing uncertainty spectrum.
Static Parameters, Dynamic Risk
Protocols like Aave, Euler, Morpho, and Compound set loan-to-value ratios, liquidation thresholds, and liquidation bonuses per asset. Risk providers like Chaos Labs and Gauntlet analyze historical volatility, liquidity depth, and stress scenarios to recommend parameters.
Once parameters are set, enforcement is static. Collateral requirements do not change as market conditions evolve. Risk may rise smoothly while constraints remain fixed, until a discrete liquidation event occurs.
This is ex ante risk calibration rather than continuous risk enforcement.
The Temporal Mismatch
Risk changes continuously. Markets move. Volatility spikes. Correlations shift. Liquidity conditions tighten.
Static parameters cannot track these changes. A position that was adequately collateralized yesterday may be dangerously undercollateralized today, but the protocol does not know until the liquidation threshold is crossed.
Governance can update parameters, but governance is slow. Parameter changes take weeks or months. When updates do occur, they can produce abrupt margin shocks, potentially triggering liquidation cascades precisely when markets are stressed.
Why This Works for Liquid Assets
For highly liquid crypto assets, static models work adequately because several conditions hold:
- Prices update continuously from deep markets
- Liquidation can happen atomically in seconds
- Exit times are measured in seconds, not days
- Historical data is abundant for volatility estimation
When liquidation is fast and reliable, static thresholds are sufficient. If a position crosses the threshold, liquidators act immediately. The system clears before losses accumulate.
Why This Fails Progressively
As assets become less liquid, less frequently priced, or harder to exit, static models fail progressively:
Pricing becomes discontinuous. Some assets trade daily, some weekly, some quarterly. Between price updates, the true value may have changed substantially, but the protocol does not know.
Liquidation becomes delayed. Exiting a position may require finding buyers, negotiating prices, or waiting for redemption windows. This takes days, weeks, or months depending on the asset.
Exit prices become uncertain. Forced sales in stressed conditions realize prices below fair value. The discount required to execute under duress is not fixed but varies with market conditions, often exceeding the haircut used for margin purposes.
Historical data becomes sparse. With fewer price observations, volatility estimation becomes uncertain. Confidence intervals widen. Parameters must be set conservatively.
The Capital Efficiency Problem
When static parameters must cover worst-case scenarios at all times, capital efficiency suffers. A 75% LTV that works for ETH becomes dangerously aggressive for less liquid assets. A safe LTV for quarterly-priced assets might be 50% or lower.
This chronic overcollateralization makes leverage expensive. Borrowers must post far more collateral than current conditions require. Capital sits idle, earning nothing, waiting to absorb losses that may never materialize.
The Systemic Risk Problem
Alternatively, protocols may set LTV aggressively to improve capital efficiency. This works during normal conditions but fails catastrophically during stress.
When multiple positions become undercollateralized simultaneously, liquidators must absorb large volumes. Market impact increases. Prices fall further. More positions become undercollateralized. The system enters a liquidation cascade.
For assets with delayed exit, the problem compounds. Positions become insolvent but cannot be liquidated quickly. Bad debt accumulates. The protocol socializes losses across depositors.
The Fundamental Limitation
The fundamental problem is not parameter calibration. Better analysis produces better parameters, but parameters remain static between updates. Risk curators like Chaos Labs produce continuously updating risk signals, but these signals inform governance rather than directly modifying margin requirements.
The enforcement path remains discrete. Measurement is sophisticated, but enforcement is primitive.
How Traditional Finance Solves This
Traditional clearinghouses do not separate risk measurement from enforcement. They operate a continuous loop:
- Measure exposure and estimate loss distribution
- Translate loss estimates into haircuts
- Enforce haircuts as collateral requirements
- Repeat continuously as conditions change
Haircuts adjust smoothly. When volatility rises, haircuts increase gradually. Positions are margined proactively rather than liquidated reactively. The system stays in balance.
This is clearing. It is how risk is managed for trillions of dollars in traditional markets. It is the infrastructure layer that DeFi is missing.
What RAVA Provides
RAVA brings clearing on-chain. It provides continuous VaR estimation that produces two related outputs: haircuts for collateral valuation and discounts for execution pricing.
Haircuts determine borrowing capacity under orderly conditions. They adjust continuously as risk changes, tightening leverage gradually rather than through discrete liquidation events.
Discounts determine execution prices during liquidation or standing bid scenarios. They include additional costs beyond the haircut: market impact, time-to-exit, adverse selection, and slippage.
Liquid assets receive tight haircuts and discounts that respond quickly to volatility changes. Less liquid assets receive wider values that account for pricing uncertainty and exit risk. The gap between haircut and discount widens as liquidity decreases.
The result is capital efficiency that adapts to conditions, systemic stability through gradual margin tightening, and reliable execution pricing when liquidation becomes necessary.