Listing gmBTC, gmETH on Aave Arbitrum
Intro:
GMX Protocol is a leading DEX offering derivatives and one of the most popular in DeFi today. The integration of dynamic velocity-based funding rates has led to the protocol maintaining a healthy distribution of long-short OI skew over time. Integrating gmBTC and gmETH as collateral assets in the Aave V3 Arbitrum Pool has the potential to create new demand for borrowable assets on Aave V3, such as BTC/ETH and stablecoins.
Main use cases:
Borrowing underlying assets (BTC/ETH) to hedge delta exposure when supplying liquidity in gmPools while generating yield. Specifically concerning risk on Aave, if the underlying debt asset (BTC/ETH) scales in price while the collateral asset (gmToken) maintains a delta of approximately .5, liquidations can occur. This is due to the gmToken being comprised of 50-50 BTC or ETH and USDC, assuming the distribution of OI is sufficiently balanced.
Borrowing stables to increase delta exposure when supplying liquidity in gmPools while generating yield. For example, supplying gmAsset as collateral, borrowing USDC, depositing single-sided USDC into gmAsset pool, and attempting to create a delta exposure of ~1. If the price of the gmToken, i.e., the underlying asset * ~.5, liquidations can occur.
Methodology:
Supply Cap:
Every GMX Market has a reserve factor that allows open interest to be capped to a percentage of the pool size. This reduces the impact of profits of short positions and reduces the risk that long positions cannot be fully paid out.
GMX has two separate parameters for reserve factor:
- open interest reserve factor - it is impossible to open a position if new OI > pool size * open interest reserve factor
- reserve factor - it is impossible to redeem gmTokens if current OI > pool size after withdrawal * reserve factor
For both BTC and ETH, the reserve factor is 95%, and the open interest reserve factor is 90%. Thus, to quantify the supply cap, we take the delta between the reserve factor and the open interest reserve factor (reserve factor (95%) - open interest reserve factor (90%)), i.e., 5% * pool size. Effectively, this means we assume the worst-case scenario concerning the gmToken’s ability to be liquidated and withdrawn based on the parameter values set. Put differently, in the scenario of a fully utilized pool encompassing both long and short positions and factoring in the open interest reserve factor; our methodology ensures that the protocol is equipped to manage the flow effectively. This indicates that even in the scenario where all positions are liquidated and collateral is recovered, the protocol can still manage its operational needs effectively. However, it’s important to recognize that not all borrowers are expected to employ identical strategies, contrary to this assumption.
Furthermore, the full utilization of the OI reserve factor on both sides suggests a lack of significant market movement in the underlying asset. Typically, in the presence of strong market fluctuations, either short or long positions on GMX would have been closed out. This observation is inversely correlated with the anticipated demand for gmBTC as collateral, implying that substantial market movements would be required for liquidation on Aave, due to the .5 delta.
Given the pool is valued in USD terms, taking the median of the pool value over the last 90 days and multiplying by our 5% margin, this comes out to 4.3m ($6m) for gmBTC and 3.5m ($4.5m) for gmETH.
BTC:
ETH:
Liquidation Bonus:
Unlike traditional liquidations, the liquidating & withdrawing of gmTokens from the pool is non-atomic. Thus, flashloans cannot be utilized to liquidate positions. In addition, there exists a scenario where liquidations may not occur if the collateral received by the liquidator is greater than the liquidity available to be withdrawn or redeemed, based on the reserve factor. Thus, we must provide a significantly larger liquidation bonus (LB) for liquidators, who are required to utilize their capital to liquidate positions and theoretically take on additional delta risk if they choose to liquidate when redeeming is unavailable.
Looking ahead, the implementation of atomic liquidations will represent a significant advancement, further diminishing the risks associated with listing gmTokens as collateral. This development will enhance the efficiency and safety of the protocol, paving the way for additional optimization of parameters.
Liquidation Threshold:
The gmToken underlying assets, i.e., BTC and ETH, are naturally listed on Aave with discrete LT/LTVs. Thus, to account for higher LBs, whereby more debt needs to be paid off to revert the position to a healthy state, following Chaos Labs LT framework, we recommend the values in the chart below.
Additional Note:
Other platforms like Abracadabra may offer gmTokens as collateral with more aggressive Loan-to-Value (LTV) and Liquidation Thresholds (LTs). In the edge case scenario presented above, if the underlying gmToken or the asset it represents experiences rapid price fluctuations, these platforms effectively gain priority in liquidations, potentially consuming a significant portion of the gmPool reserve factor buffer before Aave can react.
Recommendation:
| Parameter | gmETH | gmBTC |
|---|---|---|
| Isolation Mode | No | No |
| Borrowable | No | No |
| Collateral Enabled | Yes | Yes |
| Supply Cap | 3,500,000 gmETH | 4,300,000 gmBTC |
| Borrow Cap | N/A | N/A |
| LTV | 60% | 55% |
| LT | 65% | 60% |
| Liquidation Bonus | 12.00% | 12.00% |
| Liquidation Protocol Fee | 10.00% | 10.00% |
| Reserve Factor | N/A | N/A |
| Variable Base | N/A | N/A |
| Variable Slope1 | N/A | N/A |
| Variable Slope2 | N/A | N/A |
| Uoptimal | N/A | N/A |
| Flashloanable | No | No |
| Siloed Borrowing | No | No |
| Borrowed in Isolation | No | No |
| Emode | No | No |