Removing the Human Factor from DeFi Risk Management - TOKEN2049 Singapore 2023

Okay, thank you. Good afternoon everyone. I am yon ver. I'm here to talk about DeFi risk management . DeFi risk management is a very broad topic, and it is at the core of our activities.

This is related to financial risks , the risk activities we typically see are around smart contract bugs and hacks. However, as DeFi products and protocols become more complex , a number of financial risk factors are also coming to the fore. In this talk, I will focus primarily on lending markets. The main risk factor to consider here is the 'loan-to- factor to consider here is the 'loan-to- value' ratio, i.e.

how many American dollars you can borrow , for example, against Bitcoin or Ether collateral. This is a crucial metric in DeFi lending markets, as it not only affects protocol security and solvency , but also has a great impact on the user experience . Some information about me. I am the founder of B Protocol . It is a DeFi protocol that has been available for almost three years , aiming to reduce risks in the lending market .

I am also a main contributor to RiskDAO, which provides risk analysis services to various platforms . Previously I worked as CTO at Kyber Network, a Singapore-based decentralized exchange, where we built Wrapped Bitcoin. My journey with blockchain began in 2016 after completing my PhD in Computer Science . When Ethereum was on proof-of-work, I led the way in building the first Ethereum decentralized mining pool with other co-founders . My background is technical technical .

I have a PhD in computer science and I have a lot of technical development experience. So, when we want to reduce the risk factor , it is appropriate to say what we have been working on for the last one or two years . Recently, especially in the past year, we have been providing risk reduction services for many DeFi protocols . This desire to reduce the human factor is intended to replace ourselves with a more decentralized, smart contract-based approach . Risk management has a disreputable history with traditional finance; Some blame this for the 2008 credit crisis .

It is very biased, vague and political . In traditional risk management management , it is mostly about rating the debt of companies or sovereign countries and the quality of that debt . It is well known that this is biased against America . For example, America's debts. Well, this is the main reason for the birth of Bitcoin as an anti-financial crisis movement of 2008.

However, despite the advent of crypto , risk management is still handled by humans . We saw Three Arrow Capital go bankrupt ; Some creditors instigated the liquidation, believing it was the right business decision , but it ultimately led to complete bankruptcy. FTX triggering liquidations for Alameda is not just bad risk management, it's a major fraud. Ultimately, when the power to manage such matters is in the hands of humans, these are the results that come. So, this is something we see a lot of, not just in the traditional financial sector, but also in crypto .

But, this will never happen in DeFi, right? Because DeFi is synonymous with transparency. However, while DeFi lending markets are transparent and decentralized , they are not immune to politics and the biases of human decisions. An example of this is that many lending markets have their own third- own third- party risk vendors who provide risk management practices . Five months ago, we saw a risk vendor recommending a halt to liquidations in light of market conditions, knowing that this would lead to bad debt in other markets as well .

The complaint here is not against the lending market or against the very professional and intelligent risk vendor. But halting liquidations is, in a way, the path that led to FTX's collapse. Another recent example, not many people are paying attention to what's happening in DeFi, but a founder who took out a huge loan against his own native token has become a very hot topic. Here the gauntlet, the risk vendor, has a decision to make about what to do. Here I am reading what Mark Zeller said , he represents the Aave DAO community , he has a different opinion , looking at these two sides, you can't say which is right and which is wrong, right?

Both sides are very professional and honest, you know? And in the middle, there is an actual user using a DeFi platform , and now there is a proposal to change the rules to improve his situation . So, I repeat , I can't say for sure what is right and what is wrong here. This is not what you would expect from a DeFi protocol . This is not how Uniswap behaves.

More generally , I speak from over two years of experience in risk management in DeFi protocols. So, when you determine the loan-to-value (LTV) ratio , you have a lot of quantitative and measurable metrics in mind . However, you also need a lot of domain- you also need a lot of domain- specific knowledge . For example, assets like USDT have been very stable over the past few years . But, it is also somewhat reasonable to have a somewhat biased opinion of them, given their structure.

But, if you start analyzing the users you're working with, it gets even worse . For example, analyzing user activity is not a big deal, and if that user is most active during Asian business hours, it can be determined that he is probably from that region . Based on that, you might want to give him a bad or good risk score . So, all of this is very common in a traditional economy, but you wouldn't expect something like this from a DeFi protocol that is unbiased and unbiased . So, our vision at B Protocol and RiskDAO is to think about how we can do this differently.

We aim to replace all of these risk vendors, including ourselves, with a smart contract . Now this smart contract takes quantitative data like DEX liquidity and asset volatility as input. Getting this data online is not easy, but it is an important part of what we do. Moreover, our intention is for everything to be fully automated and objective in the future. But considering the real situation in the short term, some human intervention will have to be allowed .

So here each organization can decide some confidence factor . For example, if you have less confidence in USDT than USDC , you can specify that, or more broadly a DAO can also specify that . We will allow some opportunity for human input , but it will be a single number input . This makes everything transparent and allows users to understand the risk and confidence factors that different protocols carry . Based on this input, the smart contract provides a recommended loan-to-value ratio, along with other parameters such as interest rates, supply, and collateral cap.

This is based on a mathematical formula we developed . There are many inputs to this formula, so it is not possible to discuss them in detail here. All inputs here can be obtained from the blockchain, or read directly from the smart contract via events . There is only one parameter that is completely subjective. Based on this individual parameter , we can compare the risk of different platforms.

I hope you understand what we've done and where and how it can be used . Currently , DAO lending markets are spending a lot of money, time, and effort negotiating loan-to-value ratios . Our goal is to fully automate this process. Even in the interim, the DAO may offer some templates based on its ideas and confidence factor. Then, there will be no unnecessary discussions and arguments.

Usually, discussions and disputes occur at the most crucial times . These things happen when you are under a lot of stress , it's not the right time to discuss such things , you don't want to seem political, but you want to maintain your protocol . So now the DAO can be pre-determined and the actual decisions will be made by smart contract without any political bias . Also, listing assets is a very important role in lending markets. So here, according to these target criteria , a dove can more automatically list assets, looking at quantitative data through the risk oracle it provides .

On various assets. And to put it in a broader context, risk management is currently a very political process; It comes with a lot of manual operations, and of course manual operations are known to have bugs, for example configuration bugs, because you want to set a one-to-value ratio . You forgot to add a few zeros and users were liquidated for no reason and actually lost money. It is very governance heavy and involves human emotions, as well as being very difficult to get a hold of for community discussions. But it also takes time, every decision you make can easily take months, you know?

Finally, it is opaque . Currently, it is ruled by risk vendors who use closed source models and they come to Doe and say, "Look, this is our recommendation," vaguely explaining how they came to the decision, but there is no real way to fully understand it . With our approach, things become more neutral ; The smart contract makes the decision. It is fully automated. So, it saves a lot of effort and resources as well .

And of course, DeFi is transparent and decentralized, which is how it should be . So, if you want to know more , we do n't have time for questions, but meet me offstage or you can follow us on Twitter , contact me on Telegram , we have a website for B Protocol and Risk Do , thank you.