> ## Documentation Index
> Fetch the complete documentation index at: https://docs.webacy.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Vault Ratings
> How Webacy scores DeFi vaults on a 0–100 composite risk scale from weighted sub-scores, additive penalties, and hard state floors.
This is the **V2 risk-score methodology** — the 0–100 composite that the V3 vault surface maps to a letter grade. For the V3 framework (seven weighted categories, composite letter grade, criteria taxonomy) see [Vault Risk V3](/api-reference/vault-risk-v3/v3-overview). The 0–100 score here is the same value V3 reports as `composite.score`.
**How it works:** Every vault gets a 0–100 composite risk score (0 = safe, 100 = critical). It's built from \~20 sub-scores, each weighted by importance. Additionally, additive penalties and hard state floors can force the score higher for binary danger conditions. A higher score means a riskier vault.
Every number in the score can be traced back to a specific signal. The API exposes the full sub-score breakdown so we can tell a user "this vault scores 78/100 because redemptions are currently closed and the admin key is an EOA with no timelock" — not just a number.
## Letter grade
The 0–100 composite maps to a letter grade via the standard `grading_scheme=v2` scale (lower risk = better grade):
| Grade | Risk score | | Grade | Risk score |
| ------------------------------- | ---------- | - | ------------------------------- | ---------- |
| A+ | 0–5 | | C+ | 47–56 |
| A | 6–12 | | C | 57–66 |
| A- | 13–20 | | C- | 67–77 |
| B+ | 21–28 | | D | 78–88 |
| B | 29–37 | | F | 89–100 |
| B- | 38–46 | | | |
See the [full grading contract](/api-reference/vault-risk-v3/framework-methodology#letter-grade-mapping).
# Outputs
| Output | What it means |
| ----------------------- | -------------------------------------------------------------------------------------------------------- |
| Composite Score (0–100) | Overall risk — weighted sum of all sub-scores + penalties + floors |
| Tier | `low` / `medium` / `high` / `critical` — bucketed label for the composite |
| listing\_verdict | `safe_to_list` / `caution` / `review_required` / `do_not_list` — actionable recommendation |
| withdrawal\_state | `normal` / `constrained` / `illiquid` / `locked` / `blocked` — current exit condition |
| liquidity\_tier | `open` / `mild_stress` / `constrained` / `illiquid` / `locked` — named label for the liquidity sub-score |
| governance\_score | Separate sub-score for governance quality specifically (centralization + upgrade + code + webacy code) |
| pct\_tvl\_withdrawable | What % of the vault's TVL can actually be withdrawn right now |
| solvency\_risk | Weighted sub-score for the "vault is losing money" signals |
| liquidity\_risk | Weighted sub-score for the "users can't get out" signals |
# Sub-scores
The weight shows how much it contributes to the final composite.
### Contract and Code Quality
| Signal | Weight | What it measures |
| ---------------- | ------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Protocol Risk | 15% | Webacy's vault protocol-risk classification, Negligible → Blacklisted. The single largest weight because it captures protocol-level risk independent of the vault's own code. |
| Upgrade Risk | 10% | Is the contract a proxy? Can it be upgraded? A vault that can be upgraded without a timelock is a rug vector. Timelocks reduce this score (7-day timelock = gold standard). |
| Code Risk | 10% | Is the source code verified on-chain? How many audits does it have? Unverified = +65 sub-score because you can't audit what you can't read. |
| Webacy Code Risk | 2% | Webacy's own vulnerability scan findings — reentrancy, unchecked calls, malicious external calls. |
### Governance and Control
| Signal | Weight | What it measures |
| ------------------- | ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Centralization Risk | 12% | Who controls the vault? EOA owner (single key) = bad. Multisig = better. A 1-of-3 multisig is still weak — we score both the threshold (how many keys needed) and the ratio (how easy quorum is to reach). Also: does a single EOA hold the strategy manager role? |
| Strategy Risk | 5% | Does the vault use external strategies or leverage? More strategies = more attack surface. |
| Asset Risk | 5% | What's the underlying stablecoin? USDC/USDT/DAI = low. Algorithmic or niche stablecoins (USR, AUSD) = high. |
### Liquidity and Exit
| Signal | Weight | What it measures |
| ---------------- | ------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Closed Liquidity | 12% | Are deposits or redemptions currently paused/closed? Redemption closed = can't exit = +60 sub-score. One of the most impactful signals. |
| Utilization Rate | 10% | For lending vaults: what % of the pool is currently borrowed? Above 95% = users can't withdraw. The curve is nonlinear — 98% scores 88/100, 100% scores 97/100. |
| Looping Risk | 4% | Recursive lending (borrowing to re-deposit). Amplifies liquidation cascades. 80%+ looping = 70+ sub-score. |
| Depeg Risk | 5% | Is the vault's share price below \$1? For stablecoins, any deviation is a red flag. Near-zero = vault collapsed. |
### Performance and Market Behavior
| Signal | Weight | What it measures |
| ----------- | ------ | --------------------------------------------------------------------------------------------------- |
| TVL Outflow | 2% | Has the vault lost significant TVL recently? 50%+ drop = 100 sub-score. Signals loss of confidence. |
| Size Risk | 2% | Small TVL vaults (\<\$50k) are riskier, less battle-tested, easier to manipulate. |
| Maturity | 3% | How old is the vault? Brand new (\<5 weeks) = +40 sub-score. Time in production reduces risk. |
### How these sub-scores map to V3 categories
In V3, you'll see these V2 sub-scores regrouped into [seven weighted risk categories](/api-reference/vault-risk-v3/framework-methodology#what-each-category-measures). The mapping below is a **conceptual correspondence** — it shows where each signal appears, not the exact scoring wiring (the V3 composite is computed from its own framework, not by re-bucketing these V2 weights).
| V2 sub-score | V3 category |
| ------------------------------------------------------------------------ | ------------------------ |
| Protocol Risk, Upgrade Risk, Code Risk, Webacy Code Risk, Maturity | `smart_contract` |
| Centralization Risk, timelock & pause controls, admin-change monitoring | `operational_governance` |
| Asset Risk, Depeg Risk, Oracle quality | `asset_collateral` |
| Closed Liquidity, Utilization Rate, Looping Risk, TVL Outflow, Size Risk | `market_liquidity` |
| Strategy Risk | `counterparty` |
`hack_exploit_history` is driven by the additive penalties and risk flags below (for example, recent exploit, ownership transfer, repeated pausing) rather than a single weighted sub-score. `chain_infrastructure` is reserved and carries no V2 signal today.
For details on what each V3 category measures and representative criteria, see [What each category measures](/api-reference/vault-risk-v3/framework-methodology#what-each-category-measures).
# Oracle Type Quality
The oracle sub-score uses a **weakest-link model** — the worst oracle type found across all of the vault's underlying markets sets the floor. Oracle quality is scored by type:
| Oracle Type | Base Score | Why |
| ---------------------------------------- | ---------- | --------------------------------------------------------------------- |
| Major decentralized oracle networks | 8 | Battle-tested, multi-source, heartbeat monitoring |
| wstETH / Pendle PT wrapped oracles | 18 | Derived from rebase math — manipulation surface if underlying is thin |
| MorphoOracle / UrdOracle (single-source) | 28 | Single price source, no cross-reference |
| Unknown / unverified | 40 | Can't assess what isn't disclosed |
Additionally: any collateral token with **\<\$5M daily trading volume** raises the oracle sub-score floor to 55 (`thin_collateral_market` flag). Below that volume threshold, a well-capitalised attacker can move the price enough to overborrow against inflated collateral — the pattern behind the Mango Markets and Cream Finance exploits.
# Additive Penalties
These fire for specific dangerous combinations or events and add directly on top of the weighted composite. Multiple conditions stack independently.
### Interaction Penalties (Compound Risk)
| Condition | Penalty |
| ------------------------------------------------------------------------------ | ------- |
| High utilization (>95%) + single concentrated borrower | +10 |
| High utilization (>95%) + single concentrated depositor (holds ≥50% of shares) | +10 |
| High utilization + major TVL outflow happening simultaneously | +10 |
| Upgradeable contract + weak multisig (threshold ≤ 2) | +8 |
| Pause function present + EOA control + no meaningful timelock | +8 |
### **Governance Behavior Penalties (on-chain activity)**
| Condition | Penalty |
| ------------------------------------------------------------------- | ------------- |
| Recent contract upgrade in last 30 days | +12 |
| Recent contract upgrade in last 30 days **+ zero audits on record** | +32 (stacked) |
| Repeated pausing (3+ times in 90 days) | +10 |
| 1–2 pause events in 90 days | +5 |
| Ownership transfer in last 90 days | +8 |
### **State & Structural Penalties**
| Condition | Penalty |
| ---------------------------------------------------------- | ------- |
| Dormant vault (very low activity, not new/bot-run) | +25 |
| Market concentration >80% in single market | +10 |
| Bad debt in underlying lending markets | +15 |
| Tight liquidation buffer (\<5%) | +10 |
| Exit liquidity \<5% of TVL | +10 |
| Webacy contract risk flagged | +15 |
| Webacy deployer risk flagged | +10 |
| Oracle gap >3x (on-chain price vs market price) | +15 |
| Collateral depegged >20% | +20 |
| ERC-4626 vault used as live price oracle in lending market | +15 |
### **Yield & Liquidity Trap Penalties**
| Condition | Penalty |
| -------------------------------------------------------------------------------------- | ------- |
| Reward-dependent yield >90% of APY from emissions | +12 |
| Reward-dependent yield >70% of APY from emissions | +8 |
| Reward-dependent yield >50% of APY from emissions | +4 |
| **Yield trap**: vault locked/illiquid + reward-dependent APY >70% | +15 |
| **Shared collateral exposure**: flagged collateral token shared across multiple vaults | +10 |
# Hard State Floors (binary minimums)
These ensure a catastrophic condition can never be masked by a low smooth score. The composite cannot go *below* these values when the condition is active.
| Condition | Floor |
| ------------------------------------------------------------------- | -------------------- |
| Redemptions closed | 75 |
| Redemptions closed + utilization >95% | 80 |
| Active depeg (share price \< 0.99) | 70 |
| Oracle risk >60 + liquidation proximity >40 | 70 |
| Exchange rate spike >2% (donation attack pattern) | 70 |
| Dormant vault | 65 |
| Exchange rate crash >1% (exploit in progress / collateral collapse) | 65 |
| Yield trap active | 65 |
| Exit illiquid (\<2% withdrawable) | 60 |
| listing\_verdict = do\_not\_list | 75 (tier = critical) |
| listing\_verdict = review\_required | 50 (tier = high) |
# Computed-but-Zeroed Signals
These are calculated and exposed in the API payload but don't add to the weighted composite — they feed into additive penalties or are surfaced directly for consumers.
| Signal | What it is |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Borrower Concentration | Top borrower's share of total borrow across the vault's underlying lending markets. If one whale controls 80% of borrowing, the vault's liquidity is hostage to them. |
| Depositor Concentration | Top depositor's share of vault shares. A single whale exiting triggers a utilization spike that traps everyone else. |
| Governance Behavior | Recent on-chain governance events: contract upgrades (30d), pause events (90d), ownership transfers (90d). Structure-based scoring alone misses behavioral patterns. |
| Yield Sustainability | Ratio of emissions-based APY to total APY. >70% = yield collapses when rewards end, causing TVL flight and utilization spike. |
| Exchange Rate Velocity | Rate of change in vault share price since last checkpoint. Spike >2% = donation attack pattern (Venus wUSDM). Crash >1% = exploit in progress or collateral collapse. |
| ERC-4626 Oracle Risk | Whether the vault's convertToAssets() function is being used as a live price oracle in a lending market. Direct asset donations inflate this rate without minting shares — enabling overborrowing against inflated collateral. |
| Oracle Gap | Worst ratio between on-chain oracle price and real market price found across collateral tokens. 9x gap = oracle is stale or manipulable. |
| Exit Liquidity Ratio | What fraction of TVL is withdrawable right now vs total deposited. |
| Market Liquidity | Absolute USD liquidity available in the underlying lending markets. |
| Market Concentration | What % of TVL is in the single largest lending market. |
| Bad Debt | Realized losses in the underlying lending markets that won't be recovered. |
| Momentum | 1-month vs 3-month CAGR divergence — is yield declining recently? |
| Flow Risk | Net capital flows — large outflows signal institutional confidence loss. |
| Fee Drain | Management + performance fees. >2% management fee = meaningful yield erosion. |
| Volatility | Share price volatility over 3 months. |
| Drawdown | Worst historical drawdown from peak. |
| Return Loss | Lifetime or annualized return negative = vault is losing user capital. |
# Risk Flags
Flags are surfaced in the API alongside the score. Each maps to one or more of the conditions above.
**Blocking flags** (force do\_not\_list): unverified, redemption\_closed, dormant
**Non-blocking flags**: depeg · high\_looping\_exposure · no\_audits · eoa\_owner · pause\_capable · upgradeable · negative\_return · lockup\_7d · withdrawal\_delay · low\_tvl · new\_vault · deposit\_closed · inactive · subvault · thin\_collateral\_market · concentrated\_borrower · concentrated\_depositor · recent\_upgrade · unaudited\_upgrade · repeated\_pausing · ownership\_transfer · exchange\_rate\_spike · exchange\_rate\_crash · erc4626\_donation\_risk · reward\_dependent\_yield · yield\_trap · emergency\_deposit\_cap · shared\_collateral\_exposure