Overview
Given the current state of the Aave V2 instance, Chaos Labs conducted a detailed evaluation of the oracle configurations and risk parameters across all listed assets. The objective of this analysis is to identify assets whose oracle setups and risk parameters can be adjusted according to their specific market condition and risk profiles. Where appropriate, these changes aim to reduce potential risk exposure, simplify oracle dependencies, and support the orderly wind-down and deprecation of the V2 instance.
Stablecoin
Oracle Recommendation
For stablecoins that are used exclusively as debt assets and do not have a CAPO-imposed upper bound, we recommend hardcoding their price to $1. In addition to operational simplicity, this approach mitigates protocol risk by preventing debt inflation driven by oracle manipulations.
More specifically, these stablecoins are exposed to risks of temporary overpricing due to oracle manipulation. In such scenarios, an attacker controlling a large portion of the stablecoin supply can artificially inflate its price, causing the reported value of the debt to increase rapidly, even though the underlying economic value of the asset has not materially changed. This creates a critical risk in liquidation dynamics.
As debt is marked at an inflated price, positions will be pushed into liquidation. At this point, the liquidator/attacker, can repay debt using assets acquired at significantly lower market prices while seizing collateral valued based on manipulated oracle prices. This results in an economically unbalanced transfer of value from borrowers to liquidators. In addition to harming borrowers, this behavior can also introduce risk to the protocol. If the debt price is sufficiently inflated, liquidators can repay a small portion of the debt while seizing the entirety of the collateral. Once the manipulation reverts, the position may be left with no remaining collateral but still outstanding debt, which effectively becomes bad debt for the protocol.
Importantly, this risk is not limited to positions where the manipulated asset represents the majority of the debt. Even when the affected asset constitutes only a small portion of a multi-asset debt position, a sharp increase in its reported price can be sufficient to push the entire position below the liquidation threshold. As a result, the full collateral of the position may become subject to liquidation, despite the manipulation originating from a relatively small component of the total debt. Given these considerations, we recommend hardcoding the prices of these stablecoins that do not have a CAPO-imposed upper bound.
In addition to mitigating potential bad debt risk, this configuration does not introduce additional risk under normal market conditions. Temporary deviations, including short-term overpricing or depegging, do not pose a risk to the protocol when these assets are used solely as debt and are hardcoded. Only in scenarios where prices remain persistently deviated does the risk profile change. If the asset trades below the peg for an extended period, the impact is primarily borne by borrowers, who effectively repay more relative to market value.
If, instead, the asset remains persistently overpriced, this may introduce potential risk to the protocol. In such cases, liquidators must acquire the stablecoin at elevated market prices to repay debt accounted at $1. If the deviation becomes sufficiently large, the liquidation bonus may no longer compensate for the higher acquisition cost, weakening liquidation incentives and potentially leaving positions unresolved. However, to mitigate this risk, we also recommend to increase the liquidation bonus for collateral assets associated with these hardcoded stablecoins, ensuring that liquidation incentives remain sufficient under such conditions.
Based on the above, we recommend hardcoding the oracle price of FEI and RAI, and increasing the LB of their associated collateral to ensure sufficient liquidation incentives.
Risk Parameter Recommendation
We propose a slight reduction in the LT of TUSD. The objective is to facilitate the gradual unwinding of TUSD borrowing positions as part of the Aave V2 deprecation process, while avoiding forced immediate liquidations for a large number of existing users. The rationale for this adjustment is consistent with the reasoning behind the risk parameter changes for volatile assets outlined below.
Volatile Asset
Oracle Recommendation
We recommend fixing the price of selected volatile assets in terms of the base currency (ETH), rather than relying on market-based oracle feeds. This approach is consistent with the treatment applied to stablecoins described above, aiming to mitigate the risk of upward price manipulation and reduce protocol-level risk arising from distorted debt valuation.
At the same time, we do not recommend modifying the oracle configuration for assets that retain sufficient market depth and liquidity, as the current market depth makes sustained oracle manipulation less likely and limits the associated bad debt risk.
In addition, we propose setting these fixed asset/ETH prices at a level above their current market-implied value (e.g., ~30% higher than spot). This serves two purposes. First, it mitigates the risk of liquidation inaction arising from discrepancies between the fixed oracle price and the actual market price. If the asset trades materially above the fixed oracle level, liquidators may be forced to acquire the asset at significantly higher market prices than implied by the oracle, rendering liquidations economically unattractive. By introducing an upward buffer, this risk is reduced, helping ensure that liquidations remain executable in practice.
Second, by increasing the effective debt value, it helps push marginal positions closer to liquidation, thereby facilitating the unwinding of smaller positions and contributing to the overall acceleration of Aave V2 deprecation.
Risk Parameter Recommendation
We recommend reducing the liquidation thresholds of all assets that currently have a non-zero LT. This change is to support the gradual unwinding of remaining positions as part of the Aave V2 deprecation process, without triggering immediate liquidations for the majority of the market, allowing for the primary existing users to react to the changes.
To assess this, we simulate user-level position snapshots under progressive LT reductions using the latest full position dataset. At each reduction level, we apply small absolute LT cuts of 1% to 10% simultaneously across all collateral assets that have a non-zero LT, while holding prices and debt fixed. For each reduction level, we recompute wallet-level health factors under the shocked LT parameters and identify wallets that are newly pushed into liquidation. Newly liquidatable wallets are defined as those with a pre-shock health factor at or above 1 that fall below 1 after the LT change. We then measure the resulting newly liquidatable collateral exposure as the total collateral-enabled supply held by these wallets across all shocked assets.
The above exposure metric represents the total amount of collateral that becomes associated with wallets entering liquidation risk under the given LT reduction scenario. However, this estimation is conservative, as it accounts for the full collateral balance held by newly liquidatable wallets, rather than the smaller portion that would actually be liquidated in practice.
The results indicate that the system remains relatively stable under moderate LT reductions, but exhibits a nonlinear increase in risk beyond a certain point. When all collateral LTs are reduced by 4%, the resulting newly liquidatable collateral exposure is approximately $319K. However, increasing the reduction to 5% leads to a sharp increase in exposure to roughly $761K. This inflection suggests that the system begins to enter a more sensitive regime beyond the 4% reduction level.
Based on these results, we recommend implementing a uniform 4% LT reduction across all collateral assets on Ethereum. This level introduces a controlled amount of liquidatable collateral, facilitating more efficient system deleveraging, while preserving the stability of the vast majority of user positions. Additionally, based on the simulation framework outlined above, we have also applied corresponding LT reductions to assets with non-zero LT on Avalanche and Polygon to accelerate the deprecation of Aave V2 on these instances.
Interest Rate Model Recommendation
Reduce IRM
For assets where healthy debt is significantly smaller than stressed debt, we recommend reducing IRM to ~0% to stop further interest accrual. In such scenarios, additional interest primarily accrues on already distressed or uncollectible positions, effectively increasing the protocol’s eventual loss. Freezing debt growth is therefore a more appropriate approach during the unwind process.
The chart below shows the composition of outstanding debt across selected borrowed assets on Aave V2, split between healthy debt and stressed debt. Healthy debt refers to debt backed by positions with a HF of at least 1, while stressed debt refers to debt associated with positions where HF is below 1. For each asset, the left column presents a 100% stacked view of this composition, showing the relative share of healthy versus stressed debt. The adjacent column shows the total amount borrowed for that asset in USD.
As shown, for assets such as MANA, FEI, and BUSD, stressed debt already exceeds healthy debt by a significant margin. In these cases, continued interest accrual is unlikely to be recovered and instead increases the notional size of bad debt. Accordingly, we recommend reducing IRM to ~0% for assets where stressed debt exceeds healthy debt, as IRM no longer serves as an effective risk management tool and instead amplifies realized losses.
Other Considerations
We suggest setting the Slope2 parameter to 100% for selected assets with non-zero IRM. These assets do not fall under the condition outlined in the previous section, and therefore their overall IRM configuration can remain unchanged. However, adjusting Slope2 allows for a more controlled response under high utilization, helping to accelerate deleveraging and liquidations without pushing borrow rates to excessively high levels. This achieves a more balanced and effective risk management outcome.
We also recommend increasing the IRM of TUSD to align with the parameters proposed in our previous recommendation, while lowering its Slope2 for the reasons described above.
Additionally, as the AMPL market does not contain remaining collateralized positions, and the existing suppliers have been previously compensated through Merkl distribution here, we recommend setting its Interest Rate Curve to 0.
Finally, we recommend increasing the IRM of WPOL. Despite being a major asset on Polygon, its oracle exhibits relatively slow update characteristics, with an inferred deviation threshold of ~1% and a heartbeat of approximately 6 hours. As WPOL remains enabled as collateral, hardcoding its price is not feasible, leaving the system exposed to potential upward price deviations. To mitigate this risk, increasing IRM introduces additional pressure on outstanding borrow positions, encouraging faster deleveraging and reducing the duration of exposure under potentially stale or distorted pricing conditions.
Specification
Stablecoin Oracle Adjustment
| Asset | Instance | Recommended Oracle Value |
|---|---|---|
| FEI | Ethereum | $1 |
| RAI | Ethereum | $4 |
Volatile Asset Oracle Adjustment
| Asset | Instance | Recommended Asset Price in ETH (1e18) |
|---|---|---|
| SNX | Ethereum | 181283037000000 |
| YFI | Ethereum | 1570701733962692240 |
| xSUSHI | Ethereum | 188874382650430 |
| ENS | Ethereum | 3750500000000000 |
| BAT | Ethereum | 68399060169053 |
| MANA | Ethereum | 53454081966022 |
| DPI | Ethereum | 29597100000000000 |
| ZRX | Ethereum | 63782978634450 |
| BAL | Ethereum | 95553200014979 |
| 1INCH | Ethereum | 57057200000000 |
| KNC | Ethereum | 86975575866599 |
| ENJ | Ethereum | 12846949494839 |
| REN | Ethereum | 2242430213652 |
| CVX | Ethereum | 1062775618954092 |
| BAL | Polygon | 95553200014979 |
| GHST | Polygon | 39940215118135 |
Risk Parameters Adjustment
| Asset | Instance | Current LT | Current LB | Recommended LT | Recommended LB |
|---|---|---|---|---|---|
| USDC | Ethereum | 87.50% | 4.5% | 83.5% | 10% |
| WETH | Ethereum | 86% | 5% | 82% | 10% |
| DAI | Ethereum | 77% | 4% | 73% | 10% |
| MKR | Ethereum | 10% | 7.5% | 6% | 10% |
| stETH | Ethereum | 83% | 7% | 79% | 10% |
| TUSD | Ethereum | 65% | 10% | 61% | - |
| LINK | Ethereum | 65% | 7% | 61% | 10% |
| WBTC | Ethereum | 82% | 5% | 78% | 10% |
| AAVE | Ethereum | 73% | 7.5% | 69% | 10% |
| WBTC.e | Avalanche | 70% | 5% | 65% | 10% |
| USDC.e | Avalanche | 78% | 5% | - | 10% |
| WETH.e | Avalanche | 82.5% | 5% | 77.5% | 10% |
| DAI.e | Avalanche | 77% | 5% | - | 10% |
| WAVAX | Avalanche | 65% | 10% | 55% | - |
| AAVE.e | Avalanche | 65% | 10% | 50% | - |
| WETH | Polygon | 82.5% | 5% | 80.5% | 10% |
| WBTC | Polygon | 75% | 5% | 70% | 10% |
| USDC.e | Polygon | 84.5% | 5% | - | 10% |
| DAI | Polygon | 77% | 5% | - | 10% |
| AAVE | Polygon | 65% | 10% | 50% | - |
| WPOL | Polygon | 70% | 10% | 65% | - |
IRM Adjustment
| Asset | Instance | Current Uoptimal | Current Base | Current Slope1 | Current Slope2 | Recommended Uoptimal | Recommended Base | Recommended Slope1 | Recommended Slope2 |
|---|---|---|---|---|---|---|---|---|---|
| ENS | Ethereum | 45% | 20% | 0% | 300% | 1% | 1% | - | 0% |
| LUSD | Ethereum | 45% | 20% | 0% | 300% | 1% | 1% | - | 0% |
| FEI | Ethereum | 45% | 20% | 0% | 300% | 1% | 1% | - | 0% |
| AMPL | Ethereum | 45% | 20% | 0% | 300% | 1% | 1% | - | 0% |
| GUSD | Ethereum | 45% | 20% | 0% | 300% | 1% | 1% | - | 0% |
| KNC | Ethereum | 45% | 20% | 0% | 300% | - | - | - | 100% |
| DPI | Ethereum | 45% | 20% | 0% | 40% | - | - | - | 100% |
| UST | Ethereum | 45% | 20% | 0% | 300% | - | - | - | 100% |
| CRV | Ethereum | 45% | 20% | 0% | 300% | - | - | - | 100% |
| FRAX | Ethereum | 45% | 20% | 0% | 300% | - | - | - | 100% |
| UNI | Ethereum | 45% | 20% | 0% | 300% | - | - | - | 100% |
| TUSD | Ethereum | 1% | 1% | 0% | 0% | 45% | 20% | - | 100% |
| WPOL | Polygon | 25% | 5% | 15% | 40% | 45% | 20% | 0% | 100% |
Disclaimer
Chaos Labs has not been compensated by any third party for publishing this recommendation.
Copyright
Copyright and related rights waived via CC0