Product status: Founding Beta
Methodology: Proof & Signal Evidence-Backed Risk Methodology 0.3.0-beta
Report as of: 2026-10-09T15:51:02.870Z
| Metric | Result |
|---|---|
| Supply APY | 4.10% |
| Utilization | 94.21% |
| Available liquidity | $139,656,830 |
| Overall risk | 23/100 — MODERATE |
| Methodology confidence | 96% |
| Market Data Quality | PASS |
| Report Release Gate | PASS |
| Commercial Methodology Gate | PASS_BETA |
| Dimension | Score | Band | Confidence | Evidence status |
|---|---|---|---|---|
| Protocol | 8/100 | LOW | 94% | PASS |
| Asset | 36/100 | MODERATE | 100% | PASS |
| Liquidity | 52/100 | ELEVATED | 100% | PASS |
| Oracle | 9/100 | LOW | 100% | PASS |
| Stablecoin | 18/100 | LOW | 90% | PASS |
| Chain | 12/100 | LOW | 90% | PASS |
The strongest risk signal in this snapshot is liquidity, not protocol or oracle risk. Utilization was 94.21%, slightly above the reserve’s 94.00% optimal utilization point, while immediately available liquidity represented about 5.79% of total supplied value.
The second material area is asset exposure / caps. Supply-cap headroom was about 19.66% and borrow-cap headroom about 15.90%.
The stablecoin dimension remains relatively low on reserve backing and transparency, but the methodology explicitly assigns risk to the issuer administrative control surface because the modeled USDC contract design supports upgrade, pause and blacklist controls.
| Dimension | Factor | Score | Evidence-based explanation |
|---|---|---|---|
| Liquidity | Utilization vs optimal | 65/100 | Utilization is 94.21% versus 94.00% optimal. |
| Liquidity | Available-liquidity ratio | 65/100 | Available liquidity is 5.79% of total supplied value. |
| Stablecoin | Issuer administrative control surface | 60/100 | Upgrade, pause and blacklist controls are present in the modeled Circle EVM stablecoin design. |
| Asset | Supply-cap headroom | 40/100 | Supply-cap headroom is 19.66%. |
| Asset | Borrow-cap headroom | 40/100 | Borrow-cap headroom is 15.90%. |
| Asset | Collateral LTV | 25/100 | Max LTV is 75.00%. |
The live Aave MCP reserve-details response identified:
AaveV3Ethereum;0x87870Bca3F3fD6335C3F4ce8392D69350B4fA4E2;0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48;Chainlink;0x3f73F03aa83B2A48ed27E964eD0fDb590332095B;The v0.3 score is a transparent comparative heuristic. It is not a probability of loss, guarantee, credit rating, suitability assessment, or personalized recommendation.
A PASS_BETA commercial gate means the evidence and methodology completeness requirements passed for a low-cost informational beta report. It does not mean the model has been statistically calibrated to predict future losses.
This report measures and explains observed market risk. It does not instruct a reader to buy, sell, supply, borrow, repay, allocate a portfolio, or otherwise transact.
Decision-support principle: Measure → Explain → Simulate → Let the user choose.