Vesta Finance: System Parameterization Risk Analysis
During the month of May 2022, Risk DAO was engaged to perform a risk analysis for Vesta Finance’s system parameters. The full report is publicly available here. In this blog post, we summarise the results of our analysis and compare them to the current state of Vesta Finance.
Background
Vesta Finance is a zero interest rate lending protocol. It allows users to borrow VST, a USD-pegged stable coin, against a list of collateral assets. Similar to Liquity protocol, it allows users to stake their VST into stability pools, which are used as the first liquidity source in times of liquidations.As another layer of protection, if the stability pool is empty, liquidations are executed by distributing the liquidated user’s collateral and debt among all other borrowers in the system.
Finally, to make sure VST is traded with a value of at least $1, it allows redemption of VST in return for an underlying collateral. Stability pool losses, liquidation distribution, and redemptions have severe consequences on the user experience, however the ultimate risk of any lending platform is insolvency. Hence, in this analysis, we focus on system insolvency.
While the current implementation of Vesta Finance is inspired by Liquity’s design, the team expressed their desire to pivot into a more standard parameterized model and so throughout this report we use Compound compatible definitions and notations.
In this summary we only discuss the collateral value (a.k.a LTV) and debt ceiling values, and we refer the reader to the full report for an analysis of more parameters. The collateral value is the amount of VST a user could borrow for each $1 of deposited collateral. The debt ceiling is the maximum amount of VST that can be borrowed by all users against a certain collateral. Both values are configurable and can be set with a different value for each collateral type.
Quantitative simulation
At Risk DAO we’ve developed a novel simulation model. We take real world liquidation data of popular assets from centralised exchanges, along with the price trajectory of the assets. We then extrapolate the liquidation sizes and price trajectory to the asset we wish to analyse, and simulate the outcome based on the asset’s available DeFi liquidity. Our approach eliminates most assumptions over user behaviour during market crashes, and makes it more feasible to analyse the risk of a platform prior to its launch, and for multichain lending platforms, where the data for user behaviour is even more sparse.
The simulation model is used to reason about the safety of collateral factor and debt ceiling values (and in the full report we also use it to reason about additional risk parameters).
With the model we run hundreds of thousands of simulations to estimate the expected system insolvency under each scenario. The figure below shows a single simulation run, with price trajectory (in green), market liquidity (red), liquidation volume (yellow) and stability pool size (orange) change over time.
Qualitative assessment and best practices
Vesta Finance, and the entire Arbitrum L2 ecosystem, are both at an early stage in their life cycle. As such, there is a lot of uncertainty and potential issues that mathematical models cannot quantify. Some potential issues, such as smart contract bugs or rug pulls, are better mitigated by placing a debt ceiling to cap the worst case scenarios. Other issues, such as a lagged oracle price can be mitigated by reducing the collateral factor.
For the Vesta Finance analysis we took the following considerations into account:
- Price oracle quality: e.g., Chainlink price feed vs TWAP price oracle.
- The quality of the on-chain liquidity: e.g., is the liquidity organic or incentivised? Is it owned by the protocol? Etc.
- The collateral underlying asset smart contract complexity and other systemic risks.
Quantitative analysis results
Our quantitative simulation runs different monthly liquidation volumes, which range between x0.5 to x6 of the collateral debt ceiling. For example, if the asset debt ceiling is $10M, then we run simulations for monthly liquidation volumes that range between $5M to $60M.
Then for every liquidation volume and debt ceiling we run multiple simulations and summarise the result with a number that ranges between 0 to 1. A result of 0.7, for example, suggests that the maximal loss during the worst case liquidation that was simulated was 30%. Thus, under ideal settings, a collateral factor that would have been set to that result would bring the maximum loss to 0. In practice however, the liquidation incentive should also be taken into account, and should be subtracted from the result.
The results of the simulations also depend on some market conditions, such as the DEX liquidity and the stability pool size. We ran the simulation for tens of thousands of different market parameters, and it is possible to browse all these results here.
Below we present the results that best represent the current market conditions. These results can be used to reason about the safety of the current Vesta Finance configuration.
ETH
Currently Vesta does not have a debt ceiling for the ETH collateral, however the current VST debt that is backed by ETH amounts to $2.7M. The stability pool of ETH consists of 25% of the total debt.
The simulation results for these parameters are depicted in the figure below. As Vesta is still an early stage protocol, we recommend that an average of the results for factors 0.5–2 should be used.
Result: As mentioned previously, ETH does not have a debt ceiling in Vesta, and we advise the team to set a ceiling. If there were a debt ceiling of $10M, the current risk parameters in Vesta would successfully pass our simulation stress tests.
gOHM
Currently Vesta has a debt ceiling of $2M of gOHM. The simulation results for these parameters are depicted in the figure below.
Result: The current risk parameters in Vesta successfully pass our simulation stress tests.
renBTC
Currently Vesta does not have a debt ceiling for the renBTC collateral, however the current VST debt that is backed by renBTC amounts to $0.6M. The stability pool of renBTC consists of 10% of the total debt.
Result: Currently renBTC does not have any debt ceiling, and we advise the team to set one. For the current renBTC debt, the current risk parameters in Vesta successfully pass our simulation stress tests.
GMX
Currently Vesta defines a $1M debt ceiling for GMX collateral. The simulation results for these parameters are depicted in the figure below.
Result: the current risk parameters in Vesta successfully pass our simulation stress tests.
DPX
Currently Vesta defines a $3M debt ceiling for DPX collateral. The simulation results for these parameters are depicted in the figure below.
Result: the current risk parameters in Vesta successfully pass our simulation stress tests.
Qualitative analysis results
The table below shows the result of our qualitative analysis and in the full report we also give a score for each of the assets.
Price oracles
Vesta Finance uses a Chainlink price feed for ETH, however they currently do not use any sanity check for the return result. Furthermore, for renBTC the price of BTC is taken, without a sanity check for a possible deviation of renBTC to BTC price. The GMX oracle is a Uniswap v3 TWAP oracle, with a TWAP window of 30 minutes, and thus will lag. The DPX price oracle is operated by the DPX team.
Liquidity
renBTC on Arbitrum heavily depends on a single DEX, namely, the renBTC/WBTC pair on CurveFi, with only 200 liquidity contriubtors. gOHM and DPX liquidity are incentivized by liquidity mining and heavily fluctuate on Arbitrum.
Additional Risks
renBTC suffers from custodian risk. gOHM, GMX, and DPX can all be minted and/or upgraded upon a governance decision. More details about their governance are given in the full report.
Debt ceiling
In the full report we advise the team to apply a debt ceiling policy that stems both from the quantitative and the qualitative analysis. At the time of writing, the policy is yet to be implemented, and in particular, some of the assets still have an infinite debt ceiling.
For more details please see the full report.
