[ARC] Gauntlet Risk Parameter Recommendations (2023-03-20)

[Pauljlei]

Thanks for the question, @luigidemeo88.

@luigidemeo88

Particularly for BTC.b and WETH.e it makes little sense for assets that can be bridged and arbitraged on demand to have such aggressive supply caps. Particularly, these assets do not require merchants or special fees to redeem and withdraw, they can be easily bridged.

This decision was made after Gauntlet’s analysis of the situation and considering various factors, which are detailed below. Extreme price movements can cause insolvencies in the protocol.
Approximately ~70% of all liquidations on Avalanche immediately trade out of their position. All of the BTC.b liquidations (13 total greater than 1,000 USD) in the last month immediately trade out of their positions. See these examples 1, 2, 3. That is the basis of considering on-chain liquidity for setting borrow and supply caps at this moment. The learning from the USDC price deviation was that DEX liquidity is a NECESSARY condition for liquidations on Aave V3 Avalanche. This condition seems to hold for BTC.b as well. By lowering the caps, we aim to mitigate the risk of price manipulation attacks and protect the protocol’s growth. The cap we set relative to market caps is balanced to defend against attacks while still allowing for revenue generation.

• Volume of bridged assets vs non-bridged assets: A significant amount of bridged assets like DAI.e left the ecosystem in comparison to non-bridged assets like USDT and USDC. This indicated a heightened level of risk and warranted a more conservative approach to managing the lending protocol.
  Recall that our borrow and supply caps serve to protect the protocol from price manipulation attacks, with liquidity being a key factor in the calculations. Please note the net outflows in BTC.b below.
  
• Average time to bridge: The time taken to bridge assets is crucial in times of duress. Per the Avalanche docs, on average, it takes 1 hour to bridge to Avalanche from the Bitcoin network, and 20 minutes from Ethereum. Our aggressive recommendations take into account global liquidity, including liquidity in centralized exchanges. However, for bridged assets, this may hinder healthy liquidations from occuring during black swan events. By lowering the caps, we aim to encourage instantaneous liquidations and improve the stability of the protocol.

@luigidemeo88

In my opinion and specifically in reference to the proposed changes to BTC.b, WETH.e, DAI.e, this is a pretty draconian measure for assets that have very little to do with the USDC depegging event.

Lowering DAI.e caps was a pertinent response to the depeg, as during the event, the asset dropped 8% in price and resulted in $324K liquidated on the protocol (and a relatively small amount of insolvencies). As mentioned, in the process, a majority of the DAI.e liquidations instantaneously traded out of their positions. By reducing the caps, we aim to provide a more stable and secure lending environment, ensuring the long-term success and growth of the Avalanche Aave v3 lending protocol.

Please refer to our Supply and Borrow Cap Methodology for a detailed explanation of the factors considered in setting these caps. As this is a response to a crisis, we had opted to use conservative caps that reference only on-chain liquidity. Moreover, as outlined in this methodology, we propose slightly higher aggressive caps for some bridged assets in order to promote protocol usage and asset growth; however, this is secondary to protecting the protocol in times of volatility. We will continue to monitor liquidity conditions and make recommendations as they evolve.

A Note on LSTs

@luigidemeo88

LSDs and redeemable assets such as BTC.b cannot be viewed in the same lens as native assets.

With respect to LSTs such as sAVAX, we had not recommended changes to sAVAX post-depeg, as the liquidity had widely recovered. We detail our methodology for setting borrow and supply caps for liquid staking tokens here, and welcome further discussion in that ARFC.