Morpho vaults held over $10 billion in deposits as of March 2026. DeFi vault coverage for those positions was not available at institutional scale until Catalysis, a DeFi-native coverage provider, built its underwriting engine on Credora’s risk rating dataset. This case study explains how that works: what Credora measures at the market and vault level, how Catalysis translates those signals into coverage premiums, and what the numbers look like on a real Morpho vault.
The gap: yield is priced in real time, DeFi vault coverage is not
An allocator depositing into a DeFi vault underwrites a chain of assumptions: that collateral remains liquid under stress, that liquidation mechanics function when volatility spikes, that oracle systems stay reliable, that governance cannot shift the vault’s parameters without notice.
DeFi infrastructure prices yield in real time. It has not built equally strong infrastructure for quantifying the probability or severity of loss. Most vault comparison frameworks still rely on qualitative descriptors. Those labels are not sufficient for coverage pricing. You cannot price DeFi vault coverage without explicit loss distributions.
Credora’s rating framework addresses that gap. Each vault receives a Probability of Significant Loss (PSL): the likelihood of a market experiencing bad debt exceeding 1% of principal. PSL is the primary input Catalysis needs to price coverage.
How Credora calculates vault PSL
Credora’s vault rating methodology works in two stages: first at the market level, then at the vault level. Full methodology documentation is available here.
Market-level PSL: six risk dimensions. Every market within a vault first receives its own PSL. Credora derives this from six dimensions of collateral risk. Structural Mechanics examines protocol parameters: LLTV thresholds that set the risk boundary, LIF incentives that govern liquidation behavior, and oracle characterization that determines how asset prices are fed into the model. Asset Quality Foundation is an independent calculation of the Probability of Default for the underlying collateral, which feeds directly into loss simulations when liquidations or market shocks occur. Leverage Concentration segments active borrower positions into LTV tranches to identify how close borrowers are, in aggregate, to liquidation triggers. Tail Risk Modeling applies extreme value theory, specifically Pareto distributions, to simulate Black Swan scenarios: the low-frequency, high-severity events that matter most for coverage pricing. Predictive Behavior Analysis uses machine learning (logistic regression) to estimate the likelihood of borrower rebalancing during drawdowns, whether borrowers will add collateral or repay debt before liquidation becomes necessary. Stressed Liquidity Testing evaluates available market depth through slippage curves and real-time order book data, determining how much collateral can actually be liquidated at given prices under stress.
These six dimensions interact. DeFi losses rarely emerge from a single failure mode. A collateral asset may appear liquid under normal conditions but become illiquid precisely when a liquidation is required. Tail risk and leverage concentration tend to co-occur. Credora’s simulation framework captures how these dimensions compound rather than treating each in isolation.
Vault-level aggregation: Anchor PSL and two modifiers. Once each underlying market has a PSL, Credora computes the Anchor Vault PSL as a weighted average, with weights set by the vault’s current asset allocation. A vault allocating 60% to Market A and 40% to Market B carries an anchor PSL that reflects those proportions. Unallocated markets are incorporated to account for potential shifts in exposure.
This weighted aggregation captures a risk that per-market analysis misses. A vault concentrated in ETH-derived collateral (wstETH, rETH, cbETH) may appear diversified across markets, but a 40% ETH drawdown stresses all of them simultaneously. Vault-level aggregation captures those correlated exposures.
Two modifiers then refine the Anchor PSL. The Curator Modifier evaluates the vault manager’s track record in months, total assets under management, and number of active pools. These inputs produce a composite score that places the curator in one of three tiers: Tier 1 curators with strong experience receive a +0.25 notch improvement, Tier 2 receive no adjustment, and Tier 3 receive a -0.25 notch penalty. The Governance Modifier evaluates guardian type and timelock duration. Vaults without a dedicated guardian receive -0.5; multi-signature contracts receive 0; DAO structures receive +0.25. Timelocks below 48 hours receive -0.25; 48 to 72 hours are neutral; above 72 hours receive +0.25. The governance adjustment is the average of the two components.
To make this concrete: a vault with a multi-signature guardian (0) and a 96-hour timelock (+0.25) receives a governance adjustment of +0.125. A Tier 1 curator adds another +0.25. The combined +0.375 notch adjustment is then applied to the Anchor PSL to shift the vault toward a better implied rating. The same vault with no guardian and a 24-hour timelock would receive -0.375 instead, pushing the final rating lower than the underlying markets alone would suggest.
A vault with an experienced curator and a DAO governance structure with a long timelock can reduce its effective PSL relative to its underlying markets. A vault with a new curator and no guardian carries a higher PSL than its market composition alone would suggest.
What Catalysis adds: from PSL to coverage pricing
Credora provides the risk signal. Catalysis provides the underwriting layer. The two outputs from Credora that anchor the engine are the rating (a comparative signal indicating where a vault sits on the risk spectrum) and the loss curve, which answers not just how likely a loss event is, but how large it could be if one occurs.
Catalysis runs 100,000 Monte Carlo simulations for each vault, modeling a full year of price evolution for the underlying collateral assets. Critically, liquidations in these simulations are not modeled at theoretical fair prices. They are modeled at discounted prices reflecting market illiquidity and forced selling pressure. That distinction separates a pricing model from an academic exercise.
Coverage attaches at 1%. The deductible filters out frequent, lower-severity claims and focuses protection on the events that are hardest to self-insure: the 5-10% losses from smart contract exploits or stress-driven bad debt. Small drawdowns are painful but typically manageable. Capital impairment above 1% is a different category of event. The 1% attachment is not arbitrary: Credora defines a Significant Loss event as bad debt exceeding 1% of principal. Catalysis’s deductible sits at exactly that boundary. PSL-rated events and coverage-eligible events describe the same threshold.
Premium formula. Catalysis uses a mean-plus-volatility loading approach: the premium equals expected annual payout plus a loading coefficient (κ, a multiplier that sets how large a buffer sits above expected losses) multiplied by the standard deviation of the payout distribution. At κ = 0, premiums just cover expected losses with no margin. That is imprudent, because any year where realized losses exceed the average produces negative underwriting cash flows. At a kappa value higher than 0, premiums include a full standard deviation buffer, conservative enough for early-stage coverage or vaults with limited track records but potentially too expensive for broad adoption. A balanced κ provides adequate volatility coverage while keeping premiums competitive against the yields available on insured vaults. For a well-rated vault, expected annual payout runs roughly 1.25 basis points: attractive on paper, but insufficient as a standalone premium figure.
Capital adequacy. Premium income funds expected outcomes. Tail risk requires pre-positioned capital reserves. Catalysis sizes those reserves using Expected Shortfall, a measure that asks not “how often do losses occur?” but “how large are losses when they do?” It captures the average severity in the worst scenarios, not just the threshold at which they start. The simulation output shows a consistent pattern across well-rated vaults: most scenarios produce no loss, stress scenarios produce real losses, and extreme scenarios produce several percent of TVL. Premium income and capital reserves are not substitutes. The premium is not there to fund the worst case. It is there to make the product economically sound in ordinary years. The capital behind it absorbs the rest.
What DeFi risk ratings mean for coverage cost
It creates a direct incentive structure for vault curators. Stronger collateral policies, better oracle design, cleaner governance, and more audited smart contract code all translate into lower coverage cost. That connection between risk quality and protection cost did not exist before this framework. A curator investing in better governance or switching to higher-quality collateral markets can see that improvement priced into lower insurance costs, not just reflected in a qualitative label.
For institutional allocators, coverage premiums can be modeled into portfolio-level return calculations. Yield net of coverage cost, against quantified default probability, is a different decision from yield net of nothing, against qualitative risk labels. Credora’s ratings update dynamically as vault conditions change. Coverage pricing can reprice accordingly. The link between risk quality and cost of protection is not static.
Key Takeaway
DeFi vault coverage pricing requires explicit probability of loss inputs, not qualitative risk labels. Credora’s PSL for a vault reflects the weighted risk of its underlying markets, adjusted for curator track record and governance quality. Catalysis builds on that signal to produce actuarially grounded premiums with capital reserves sized to the tail. A-tier vaults on Morpho typically cost around 25 basis points annually to insure. The pricing adjusts as ratings change.
Frequently Asked Questions
How does Credora’s PSL affect DeFi vault coverage premiums? PSL (Probability of Significant Loss) is the primary input into Catalysis’s premium pricing engine. Lower PSL vaults qualify for standard loading and lower annual premiums, typically around 25 basis points for A-tier vaults. Higher PSL vaults require larger volatility buffers, raising the cost of coverage or potentially excluding the vault from the standard framework entirely.
Why does vault coverage attach at a 1% deductible? The 1% attachment point filters out frequent, smaller drawdowns that are manageable through normal operations. Coverage is designed for rare, severe events: losses of 5-10% caused by exploits or stress-driven bad debt. The deductible concentrates protection where capital impairment risk is highest and keeps premiums economically viable.
How does Credora assess correlated collateral risk in Morpho vaults? Credora aggregates market-level PSLs into a vault-level Anchor PSL using allocation weights. A vault holding wstETH, rETH, and cbETH may look diversified per-market, but all three stress simultaneously in a broad ETH drawdown. Vault-level aggregation captures that correlation; per-market analysis does not. Credora then runs 100,000 Monte Carlo simulations on top of that signal to model realistic liquidation dynamics under stress.
This case study was prepared by Catalysis Labs & Credora. Catalysis is an onchain coverage provider building rating-driven underwriting for DeFi vaults. Credora is an independent risk ratings provider and a partner of RedStone.
