Grace Sentinel Recommendations
Gauntlet and Chaos Labs provide the following recommendations for the use of the Grace Sentinel via the Aave Guardian:
| Asset | Grace Period Enabled | Duration |
|---|---|---|
| WETH | YES | 3 Hours |
Below are the independent analyses made by both parties to support the recommendations:
Gauntlet Analysis
Assuming market conditions stay similar upon sentinel activation at 10:00 UTC on 2023/11/13, Gauntlet recommends enabling WETH for the liquidation grace sentinel for 3 hours. This could offer a good balance between allowing the most significantly impacted accounts by the pause to replenish collateral, while mitigating insolvency risk.
Comment
- The current accounts impacted by the pause, with HF < 1, may contribute to roughly ~$60k in liquidatable borrows should unpause occur. These accounts mostly borrow WETH.
- Should market move by 2 standard deviations upwards from now until grace sentinel activation, impacted accounts may contribute roughly ~$180k towards liquidatable borrows, with around ~$1k in new insolvencies.
- Applying our published framework (5x std move upwards during grace period) to 3 hours shows ~$140k in additional liquidated borrows and ~$2k in new insolvencies during the grace period, assuming no refills or new positions added.
- Market rolling volatility is generally close to historical mean.
With the grace period, the primary risk to insolvency is tail asset appreciation, given large upwards moves on 2023-11-10 and 2023-11-11. Total tail borrow unable to be liquidated with the grace period remains rather small ($5k total borrow).
Simulations
Our previous analysis is generalizable to various lengths of grace period. Taking a conservative approach under our published framework and considering 5x std market moves over a 3 hour period, we find the following. Assuming no position refill and no new positions -
| grace period length | move direction | magnitude | new insolvencies | liquidatable |
|---|---|---|---|---|
| 3 | up | 5 | 2k | 200k |
| 3 | down | 5 | 0k | 35k |
| 3 | no change | 0 | 0k | 60k |
Risk Vectors during Grace Period
Under the grace period, the largest vectors that contribute to potential liquidations and insolvencies include -
- Tail asset appreciation during grace period
- Continued ZRX appreciation may cause this account that borrows $5k ZRX against $5.6k WETH to accrue bad debt.
- WETH collateralized WBTC borrow discussed previously likely remains a potential source for large liquidation volume.
- Account currently has HF of 1.08.
- Accruing bad debt is unlikely however, as BTC/ETH would need to appreciate by ~25%.
Chaos Analysis
Summary
Based on our analysis below, we recommend implementing a 3 hour grace period to balance user experience and risk management for the protocol. This duration should allow users to protect their position from liquidation, with historical data indicating there is minimal risk of significant price swings leading to bad debt within this timeframe.
We recommend implementing the grace period for WETH, which covers the majority of the impacted positions.
The main positions currently eligible for liquidation over $5K following the pause are:
- 0x8f6d072ca42432692de973b788e89794d7db1037
- 0xb87736c9c2ddbbab1d6b59955a2d5afb8713c2db
- 0x6af11ed2008131c091ba3171bd0bb3b5248316bc
- 0x43a938d7aa8cfca1734770d5b1fce3b661d89cdd
Analysis
In our analysis, we utilized various price trajectories of V2 Ethereum assets to simulate the potential bad debt to the protocol assuming disabled liquidations. The analysis assumes that no extraordinary event occurs, such as a stablecoin depeg or significant price divergences occur between WETH and stETH.
Below, we highlight the scenarios that could lead to the accumulation of bad debt in the protocol:
1. Increase in WETH Price
The protocol will start accruing bad debt at an ETH price of ~$2,410, which represents a 16% increase at the time of writing. This is caused by the following account.
The protocol will start incurring bad debt higher than $30K at an ETH price of ~$2,615, which represents a 26% increase at the time of writing.
Analyzing data from the past 12-months, this scenario is highly unlikely to occur in the 3-hour timeframe.
Supporting Data WETH:
- Maximum Daily Change - The largest single-day percentage increase observed: ~12.55%.
- Average Daily Change - The average of all daily percentage increases: ~3.27%.
- Standard Deviation of Daily Changes - Measures the variation in daily price changes: ~2.46%.
The graph above illustrates the daily price change of WETH over the past 12-months.
The histogram provides a frequency distribution of the daily percentage changes for WETH.
2. Increase in CRV Price
The protocol will start accruing bad debt at CRV price of $0.865, which represents a 32% increase at the time of writing. This is caused by the following account.
Analyzing data from the past 12-months, this scenario is unlikely to occur in the 3-hour timeframe.
Supporting Data
- Maximum Daily Change - The largest single-day percentage increase observed: ~26.4%.
- Average Daily Change - The average of all daily percentage increases: ~3.27%.
- Standard Deviation of Daily Changes - Measures the variation in daily price changes: ~4.45%.
The graph above illustrates the daily price change of CRV over the past 12-months.
The histogram provides a frequency distribution of the daily percentage changes for CRV.
3. Extreme Price Divergence Between BTC and WETH
We observed that the primary risk of bad debt to the protocol could be caused by this account becoming insolvent.
Highlighted Account at Risk: This screenshot from the Chaos Labs platform showcases an account potentially vulnerable to ETH/BTC volatility.
An increase in BTC/WETH of over 25% would trigger the initiation of bad debt accumulation. However, the probability of such a price divergence occurring within a 3-hour is extremely unlikely, as evident in the data provided below. In addition, the user has a history of maintaining a healthy position, so should this behavior continue, allowing additional time to protect the position could reduce the risk of liquidation.
Supporting Data BTC-ETH:
- Maximum Daily Increase - The largest single-day percentage increase observed: ~3.92%.
- Average Daily Increase - The average of all daily percentage increases: ~0.092%.
- Standard Deviation of Daily Changes - Measures the variation in daily price changes: ~1.30%.
The graph above illustrates the daily price change of the BTC-ETH pair over the past 12-months.
The histogram provides a frequency distribution of the daily percentage changes for the BTC-ETH pair.
Conclusion
In essence, while extended periods without liquidations carry bad debt risk, we recognize the importance of a positive user experience. Analyzing the current situation, the risk of an extra 3 hours seems minimal following the prolonged market freeze. Considering its positive impact on overall protocol health and user experience for Aave users, we support a 3-hour grace period for liquidations after markets unfreeze.