> For the complete documentation index, see [llms.txt](https://maxim-speransky.gitbook.io/doc/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://maxim-speransky.gitbook.io/doc/gitbook-risk-framework.md). # Risk Framework This framework determines whether a Morpho lending position is eligible for a vault, which strategy may hold it, how much capital may be allocated to it, and which conditions require its reduction or removal. It is designed from the lender and vault-curator perspective. It evaluates potential loss of supplied capital arising from collateral failure, market configuration, liquidation execution, and portfolio exposure. It does not assess borrower profitability or predict token returns. The framework assumes Morpho as the underlying lending infrastructure. Protocol-level smart-contract risk is assessed separately and is not re-scored for each market admitted to a vault. ## Architecture The framework contains two independent scoring groups inside a broader decision process. Asset and Market risks receive their own assessments; the remaining stages translate those results into portfolio constraints and keep the decisions current. | Component | Question | Output | | --------------------------------------------------------------------------------- | ---------------------------------------------------- | ------------------------------------------------------- | | [**Asset Scoring**](/doc/gitbook-risk-framework/gitbook-asset-scoring.md) | Is the collateral intrinsically acceptable? | Asset eligibility + Asset Score | | [**Market Scoring**](/doc/gitbook-risk-framework/gitbook-market-scoring.md) | Can this market price, liquidate, and settle safely? | Market eligibility + Market Score + Subordination Class | | **Strategy Policy** | Which mandate may hold the position? | Core / Enhanced Yield / Excluded | | [**Allocation Policy**](/doc/gitbook-risk-framework/gitbook-allocation-policy.md) | How much exposure is permitted? | Position and dependency caps | | **Monitoring** | Does the position remain eligible? | Keep / Reduce / Exit / Reassess | The framework produces a decision vector rather than one universal risk number. A representative result may read: ``` Asset: Prime Market: Robust Subordination: Material Strategy: Enhanced Yield Cap class: C2 Monitoring: Elevated ``` The labels and cap class above illustrate the output structure; the detailed policies define their actual thresholds. ## The two scoring groups ### Asset Scoring Asset Scoring evaluates the collateral independently of the lending market in which it is used. It covers the asset's issuer and control structure, credit characteristics, and operational record. Its outputs are an eligibility decision and an Asset Score. An Asset knockout ends the assessment. A market configuration, low LLTV, or favorable yield cannot make excluded collateral acceptable. Detailed definitions, knockout rules, and evidence requirements are documented in [Asset Scoring](/doc/gitbook-risk-framework/gitbook-asset-scoring.md). ### Market Scoring Market Scoring evaluates a specific configuration of collateral asset, loan asset, oracle, LLTV, interest-rate model, liquidation routes, and lender exit. Its outputs are an eligibility decision, a Market Score, a separate Subordination Class, and Confidence. A Market knockout excludes the position even when the collateral itself is eligible. A strong Asset Score cannot compensate for an oracle or liquidation path that fails the minimum standard. Detailed definitions, domain boundaries, and evidence principles are documented in [Market Scoring](/doc/gitbook-risk-framework/gitbook-market-scoring.md). Its liquidation-capacity evidence is further defined by [Liquidation Resilience Assessment](/doc/gitbook-risk-framework/gitbook-market-scoring/gitbook-slcr-methodology.md). ## Decision pipeline Every candidate position passes through the same sequence: ``` Asset knockout → Asset Score ↓ Market knockout → Market Score ↓ Subordination Class ↓ Strategy eligibility ↓ Exposure caps ↓ Continuous monitoring ↺ ``` ### Knockout before scoring Asset and Market knockouts are evaluated before their respective scores. A knockout is a hard eligibility failure, not a penalty that can be averaged away. ### Independent scoring Only candidates that survive the relevant knockout receive a score. Asset and Market Scores are kept separate and must each satisfy their own threshold. They rank eligible candidates inside their own domains; they do not compensate for one another. ### Subordination Class Subordination Class describes how much loss absorption the deployed collateralization provides after oracle, liquidation, settlement, and stress assumptions are considered. It is an ordinal position-level output, not points added to the Asset Score. It cannot rescue an excluded asset or market. ### Strategy eligibility Passing assessments establish whether risk may be taken. Strategy Policy determines where that risk may be held: * **Core** requires the highest asset and market standards and does not pursue yield through a material increase in credit, liquidity, oracle, maturity, or structural-complexity risk. * **Enhanced Yield** may accept explicitly identified additional risk when the position still clears all hard exclusions, earns an adequate yield premium, receives a defined cap, and is subject to appropriate monitoring. * **Excluded** applies when any mandatory Asset or Market condition fails. Enhanced Yield is not a residual category for positions that fail Core. It has its own minimum eligibility standard. ### Exposure caps Eligibility decides whether a risk may be taken. Caps decide how much of an eligible risk may be taken. A smaller cap cannot make an otherwise ineligible position acceptable. Caps reflect the maximum tolerable exposure to a position and to shared dependencies. Relevant inputs include strategy mandate, realizable liquidity, collateral and borrower concentration, oracle dependencies, correlated markets, maturity concentration, total vault capital, and stressed exit capacity. Exact cap rules belong to the [Allocation Policy](/doc/gitbook-risk-framework/gitbook-allocation-policy.md). ## Composition rules The following rules govern how the stages interact: * Exclusion at any Asset or Market scoring gate dominates every later result. * Asset and Market Scores remain independent; there is no cross-score averaging. * Scores rank candidates that have already passed their knockout conditions. * Subordination Class describes position-level protection, not underlying asset quality. * Strategy Policy determines where an eligible position may be held. * [Allocation Policy](/doc/gitbook-risk-framework/gitbook-allocation-policy.md) determines the maximum permitted exposure. * Monitoring may revoke eligibility, lower a cap, or trigger a full reassessment. ## Governance and discretion Risk curation and capital allocation are separate functions. Methodology rules determine which risks are eligible, while allocation decisions operate only inside those boundaries. * Hard exclusions cannot be overridden by discretionary judgment. * Qualitative judgment may reject a position, assign a lower cap, or require stronger monitoring. * Increasing risk requires reassessment against the current methodology and data. * Reducing exposure may be executed immediately when monitoring detects deterioration. * Methodology changes, assessment changes, and allocation changes are recorded as distinct events. ## Continuous monitoring Eligibility is conditional, not permanent. Changes in collateral quality, oracle behavior, liquidity, concentration, utilization, governance, or dependencies can alter any prior decision. Monitoring feeds those observations back into the relevant assessment and results in one of four actions: keep, reduce, exit, or reassess. ## Model limitations * Scores are ordinal risk classifications, not probabilities of default or expected-loss estimates. * Weights and thresholds are policy parameters informed by evidence and analyst judgment; they are not statistically fitted unless explicitly stated. * Assessments are point-in-time and depend on the availability, quality, and freshness of data. * Liquidity observed under normal conditions may disappear during correlated stress. * Audits reduce uncertainty but do not establish smart-contract safety. * Historical survival does not capture newly introduced code, governance, or dependency risk. * Subordination Class estimates loss absorption; it does not guarantee successful liquidation. * Correlated exposures may produce portfolio losses larger than individual market assessments imply.