Proposal: Higher Loan LTVs

Author: RociFi

Submission date: Apr 25, 2023

TL;DR

Temperature check for a simple pilot to assess if higher loan-to-value ratios on Venus would increase borrowing demand.

Opportunity

Many borrowers have expressed interest in higher loan-to-value (LTV), i.e. more capital-efficient, loans. For some borrowers, capital-efficiency is a differentiator that would make them switch from other protocols to Venus.

Capital-efficiency can be achieved by issuing loans with higher Collateral Factors and therefore, higher loan sizes, to borrowers who are less likely to be liquidated based on their historical DeFi activity.

Venus offering higher LTV loans can serve existing and new users more effectively while increasing revenue.

Solution

A potential pilot, Venus users can be offered higher capital efficiency loans on a user by user basis per their RociFi credit score. This increased efficiency benefits users with customized LTVs, generates more revenue to Venus, and isolates risk away from Venus.*

  • Risk of increased LTVs, i.e. possible bad debt accumulation in liquidation spirals, is segmented to RociFi liquidity pools separate from core Venus.

About RociFi Scores

RociFi enables capital efficient lending via on-chain credit scores. RociFi credit scores (36K+) have been battle-tested on our protocol across 6000+ capital efficient loans issued within the bear market with zero bad debt from price based liquidations. RociFi’s goal is to expand this utility as a public good across DeFi as a capital efficiency layer with Venus being our primary hub.

Scores are calculated using on-chain transactional and DeFi protocols usage data across most popular EVM-compatible chains.

Deep Dive into RociFi Scoring Semantics, with examples of how various addresses are scored: Semantics of RociFi Scoring. Introduction | by RociFi | On-Chain Scores & Capital-Efficient Loans | Apr, 2023 | Medium

Capital Efficiency Examples

  1. The RociFi team has completed full simulation reports for Radiant Capital and Moonwell, previously, showing strong results.
  1. User level risk monitoring

Motivation:

Capital efficiency and risk management are at the core of all DeFi lending platforms. Simply put, if protocols have better information, they can make better decisions, which can generate higher revenue with less risk. The ability to safely increase capital efficiency should be a common good across all of DeFi with Venus being a primary hub.

Key terms

Credit scores: numerical scores representing a user’s (wallet address) creditworthiness and trustworthiness based upon their on-chain transaction history. The scale is 1 to 10 with 1 being the lowest risk (best score) and 10 being the highest risk (worst score). The scores are calculated looking at numerous DeFi protocols across 9 blockchains - not simply Venus.

Conclusion

We’d like to gauge interest before a formal vote. Please leave your thoughts in the comments!

We are open to feedback or questions regarding the idea. Thanks!

6 Likes

Thanks for submitting this proposal.

I think that increased capital efficiency is a top priority to attract more users and TVL to Venus. But I understand that this can’t be done in detriment of overall security.

I get the credit scoring part of the proposal and I believe it could work really well due to all the availabe on-chain data. What I don’t fully understand is how the liquidity pools for RociFi would be isolated or separated from the core Venus LPs.

Q: Can you please explain this mechanics in detail? How would RociFi pools be separated from core Venus?

3 Likes

I was researching the posiblilty of developing a on chain, credit based risk management system for the Venus last year but gave up as it was a huge undertake. This means I have done some reseach in this field.
The current risk management system venus and others use are off-chain and rely on historical data. This approach is outdated and does not provide adequate protection for users. Banks and CeFi platforms are far ahead of DeFi in terms of risk management. They have the infrastructure, resources and time to have developed sophisticated risk management systems that are constantly updated with real-time data. If DeFi does not adopt these same technologies, it will be at a significant disadvantage once CeFi platforms enter the DeFi space.
Current risk management systems are reactive, rather than proactive. They tell us what we could have done to avoid losing money after it has already happened. This is not helpful. We need a risk management system that can predict and prevent losses before they occur. The idea of providing glorified risk management for the most basic form of finance, a collateralized loan, is half-baked. Collateralized loans are simple financial instruments that have been around for centuries. They are relatively low-risk and easy to understand.
The best way to achieve this is to develop a risk management system that is on-chain. This would allow us to access real-time data and make dynamic changes to risk parameters based on the market and user credit scores.
I am excited about the potential of this project and I believe that it has the potential to revolutionize DeFi risk management. I would love for Venus to be amonst the first to join you and innvoate. I love the concept, I love the idea behind it. Please contintue to share updates and progess with us in the community.

Bellow is just a very rough mock-up that I never completed last year. Protocol risk score and indivudal risk score limits could be set by goverance and reacts to the market. If these people can build the tech for this on-chain real time dynamic risk management oracle and dashboard we would instantly be the leading, most safe and secure platform!

Thanks for the comment @JR-ARG!

We’re leaving this part intentionally flexible so that Venus Engineering team can make the final decision on a design mechanism that makes most sense from time and security perspective.

But, in theory, the mechanics will be 2 parts:

  1. A user requests a higher LTV loan, their credit score is checked, commensurate LTV is calculated, and conveyed to the user via the front-end. The user then locks the required collateral to execute the loan. The user should not be exposed to the backend mechanics.

  2. On the backend. A separate liquidity pool will be deployed that operates as the “additional LTV” capital pool; separate from Venus’ main liquidity pool. When the user executes the borrow function, the smart contract will call both liquidity pools and deploy the commensurate capital based upon the LTV requirements. Example: If base LTV is 80% and the borrower receives 85% LTV on $100 USDC loan, Venus’ master pool issues $80 and the junior pool issues $5.

This setup allows the isolated pool to operate as the ‘junior’ tranche where higher APRs can be charged but additional risk is held outside the main liquidity pool. Also, the entire user experience should remain the same besides minting credit scores and selecting higher LTVs.

Again, we’re leaving this area of the proposal intentionally flexible for Venus Engineering and community to drive the direction that makes most sense. We stand willing and able to support in any direction. The biggest thing we’re looking to assess is demand for higher LTVs, which you believe is useful!

3 Likes

Thanks for your quick reply. I think I understand the mechanics a little better now. I’ll be looking forward to read the comments and questions from the community and eventually, an AMA with the team if it happens.

1 Like

Thank you @Isitbathtime for this great commentary! We completely agree with you that on-chain risk analytics for DeFi is still in it’s infancy; especially assessing user level risk.

We have been working on this problem since May 2021 and have built a data and analytics infrastructure that exceeds 4 billion records across 9 different blockchains; designed to clear 100 requests per second in real-time.

We’re very excited to bring this vision to life with Venus!

1 Like

Thank you for offering to bring this to Venus & appraching us with something you are willing to work with the Venus team on, I hope you get tons of feedback and we can see the creation of something great!

1 Like

I personally agree with increasing LTV so that more money can be borrowed, but at the same time the agreement may be more prone to bad debts

Thanks for the comment @moon!

This exact reason is why we’re proposing an innovative approach to increasing LTV, i.e. by segmenting the increased LTVs to a separate pool outside Venus’ main pool. That way, if bad debts do accrue, there is no systemic risk exposure to Venus. In exchange, depositors receive higher APRs for depositing into this liquidity pool for the commensurate risk. RociFi will also look to ensure the safety of these depositors’ capital by token incentives and loss reserve, at later stages.

In summary, any bad debt will not affect Venus. We’ve structured it so that the Venus community gets all the upside with none of the downside.

2 Likes

Great to see community engagement around risk management! We reviewed the proposal and want to help facilitate the discussion by providing our observations, specifically as it relates to risks to the Venus protocol and community.

RociFi notes that their credit scores are reliant on historical wallet data. We suspect that using historical data alone isn’t strong enough to set parameters in a way that protects against unexpected market volatility and irrational user behavior. For context, the proposed credit scoring (and additional cost center) would not have protected Venus from the recent price manipulation series of attacks. Furthermore, this would not have protected Venus from prior economic shortfall events. We’d like to better understand what RociFi’s involvement and decision framework would be in a situation where a new attack vector was present. For example, would RociFi actively lower LTVs to safe levels in potential shortfall scenarios? In line with this, would RociFi write the proposal and deliver the payload for this? Proposing increases to LTV needs to include quantifying trade-offs with increased risk. Value at Risk is currently used to demonstrate this to the community; however, it is unclear what metrics will be used for any proposed increased LTVs related to this proposal. Clarity around this methodology would be appreciated!

It also seems like this mechanism design can be easily manipulated. What are the security measures in place to prevent a potential exploiter from “faking” activity to attain a higher credit score? In theory, a high credit score would let an exploiter get permission to a high LTV, which directly increases the profitability of a price manipulation attack. Credit scoring generally assumes that all actors using the protocol are rational. As we’ve seen repeatedly, there are plenty of irrational actors in DeFi. It’s important to consider both rational and irrational behaviors when determining the optimal parameters for assets on Venus. How would you know that the user isn’t a market manipulator, or doesn’t decide one day to become one? All it takes is one bad actor. However, it is interesting to see that this will be conducted in separate pools and thus decrease the risk of new bad debt to the Venus core pool. How would this work exactly? Could you walk through a theoretical example of a market downturn or market risk event (like a nonparity of a stablecoin?

As proposed, if we understand this correctly, this would appear to take TVL away from Venus’s core pool. Additionally, it seems like this proposes shifting Venus’s engineering resources away from their current priorities. Can you please break down the engineering lift to implement this mechanism change? This will allow the community to properly evaluate the potential reward relative to Venus’s other priorities and growth opportunities. Our understanding is there are several other competing priorities, such as Isolated Pools, RWA, and Venus V4. The community should consider how they factor into deciding development prioritization with regards to potentially having to delay roadmap items should this proposal pass.

We understand the importance of capital efficiency to drive and share the community’s focus on it. We also want to make sure it is done in a safe and calculated manner that allows all users to understand their risks so that desires for user and revenue growth do not backfire on Venus in the long run. Looking forward to learning more about this opportunity!

2 Likes

Thanks for the comment @jakeaujus!

Re> Parameter Setting

First, higher LTV loans should only be offered to quality credit scores, i.e. addresses with the highest probability of not being liquidated, to minimize risk. We are not advocating for increasing LTVs across the board for all borrowers.

Second, we agree with your assessment that max LTV thresholds for the ‘junior’ pool should be a function of the current params in the ‘master’ pool set by Gauntlet using VaR and potential attack vectors. Junior params will be adjusted dynamically based on market conditions or distribution of scores in the junior pool. For example, if 90% of the pool has the best credit score, keeping current LTV params remain optimal. However, if the composition worsens, LTV ca be adjusted more conservatively out of an abundance of caution. When possible, collaborating with Gauntlet on setting params in the junior pool will be welcomed. When impossible, we will set clear guidance and quantify risk for the LTV selection; including crafting the proposal. If you read our simulation articles, there are some metrics included there.

Third, I want to make it very clear that I view this proposal as complimentary to the current work that Gauntlet provides rather than competitive. Credit risk scores offer a different lens to view protocol risk, i.e. user level basis (micro), whereas Gauntlet does an exceptional job managing risk from the macro level. We view the combination of micro and macro as setting the new standard of risk management in the industry, making Venus the most security conscious protocol in DeFi.

Re > Manipulation and Design

First, there are many things possible in theory, but in practice don’t work. We have issued 6000 loans with the lowest collateral ratios in DeFi and your assertion that addresses can easily fake their way to high credit scores doesn’t hold. Furthermore, we view the behavior of DeFi actors as ‘rational’ because that is what the design or incentives allow in DeFi. So, on-chain credit scores have to be reflective of this fact, which is why there are ‘trust’ features built into our credit score. Examples include has this address ever been interacted with known phishing, hacks, exploits, market manipulations, etc. Additionally, we also have a model that scores the likelihood that a particular address is malicious but hasn’t been labeled as such yet, which is also a component of the credit score. False positives do exist, but given the dangers present in DeFi, we always err on the side of caution.

Second, as you’ve correctly noted, that even if a bad actor gains access to a higher over-collateralized LTV loan, we’ve structured things to shield Venus core from any bad debt accrual. The specifics are still being discussed as to minimize strain on Engineering and time to deployment. In it’s base form, you can think of it as connected smart contracts that communicate with one another about LTV, liquidation thresholds, etc. but execute them independently based on their own logic.

Last, additional risk mitigation can be taken at first by limiting higher LTV loans to only highly liquid blue chip assets which are less susceptible to price manipulations; including sufficient liquidity to avoid liquidation cascades accruing bad debt; even in de-pegging events. Of course, it’ll be up to the community to decide the direction on which assets to start with. Again, marrying the micro and macro, coupled with highly liquid blue chips and junior pool isolation, minimize risk demonstrably.

Re > TVL and competing priorities

TVL will spread between Venus core and junior pool But, all liquidity still belongs to Venus. RociFi acts as capital efficiency engine, nothing more.

We haven’t yet defined the specs of the pilot, so Engineering resources may or may not be shifted from Venus. Regardless, the RociFi team stands ready to help as needed to defray Engineering time and resource allocation as much as possible. We built the RociFi platform from scratch to work around the limitations of Compound / Aave forks, so we have deep expertise which can reduce time and resources required to deploy.

Once the specs are finalized with Venus Engineering team, we can provide the community a proper breakdown of Engineering lift required to bring this plan to life.

We also stand ready to offer a full simulation of potential revenue gained from enhanced LTVs, if desired by the community, so they can better understand any trade-offs in monetary terms to pursue this high growth opportunity.

Conclusion
We view this collaboration between RociFi and Venus as much about risk management as it is about higher revenue from capital efficiency. Seeing more data and risk analytics from a different vantage point, i.e. user level risk scores, can only strengthen the security and risk profile of Venus for the long run. Especially, when combined with the macro risk assessments from Gauntlet. We look forward to working with you if this proposal passes!

2 Likes

This definitely seems like a great opportunity untapped, it’s like a wide-open space we can capture.

providing higher loan to value ratio would definitely get us more borrowers driving growth.

2 Likes