Don't panic, the most you can lose is EVERYTHING | Yaron Velner , B.Protocol | ETHTaipei 2023

Disneyland can someone play some Disney music hello everyone welcome back to the D5 stage and from now on until 6 PM we invited lots of um D5 protocols like very famous ones and this afternoon we'll be all focusing about derivatives and risk management so let's welcome our first Speaker uh yaron from be protocol and he'll be talking about the new risk Oracle yeah okay [Applause] [Music] yeah okay so good afternoon everyone and thank you for coming back from lunch so yeah I will talk about a new activity we have in be protocol which is about removing the human factor from risk management which is right now very human intense work in most of the fine uh so yeah what is risk management in in general you know so there are a lot of risk aspects to finance in in D5 people people are still very much focused on the smart contact security risk but here the focus is more about economic risk especially in the traditional worlds it's about a defaults of loans in the D5 world is it is around setting certain parameters and LTV values to land in Market so in a traditional World typically you need to wait you know how risky alone is how is it safe to give a loan to a country an institution Etc yeah yeah and already in the traditional world this is very notorious [Music] known to have a lot of political aspects in it American agencies tend to give you know very strong rating to American institutions to American government the 2008 subram crisis allegedly was contributed a lot by bad ranking which is not was not only you know some kind of calculation mistake but some would argue you know it was biased from political opinions and a lot of human factors into it and of course you know from the 2008 crisis Bitcoin and cryptocame but of course risk management in crypto still suffers or at least potentially could suffer from the same issues a recent bank up say we saw for example it was reported that Genesis trading didn't liquidate three Aero capital loans because you know it had some hope that the price of a grayscale Bitcoin will go up and eventually error will become solvent again FDX you know post the liquidation to Alameda this is of course you know outright fraud was not just bad risk management but again once and you know maybe the three error story is more reflecting you know you open the door to humans to get these decisions very often these decisions make a lot of sense but once the risk is materialized you know there's no way back and then humans being humans try to cover for it and often end up in a better result this happens not only in crypto there was a recent incident with the London Metal Exchange but this can you know typically happens in you know most of the exchanges at certain point they decide to freeze the market as a byproduct of it some of the big Traders there was not got liquidated you know we will never know what was the true cause for causing it was it really political or some insolvency issues but this problem definitely arrives also in a traditional Finance as well and then you would expect okay but Defy is all about transparency this would never happen in defy you know all the major defy Landing markets are very much solvent so this is an example from the last USD CD Peg uh this is from other form so this is a you know something Gangland recommended but this is definitely not something again gunland the recommendation you know made perfect sense uh but so here you know community members for this purpose recommended to pose the protocol and stop liquidations because of the usdc debugging and you know fully aware that you know also other positions who didn't care about depth and yeah you know like as a community member this is perfectly something reasonable to do the upsides downsides from a trading perspective it might was you know like the best trading decision factoring all the risk and reward and Gauntlet are definitely smart people also pretty much you know the kind of a cell party service so also no hidden agenda but eventually this is something you know humans have to decide on the Dao has to decide on uh while Gauntlet might be perfectly objective of course every token or every other token holder has their own agenda their own incentives which might be inside or outside other itself and the outcome eventually can be again insolvency when you freeze liquidations you know so ideal things get better or get worse later it can go both ways and okay and even in D5 we see that eventually human and people endows are the one who are taking the decisions so some background on myself so I'm the founder of beep protocol I'm also contributor to the risk Dao in the rest now we do give a risk analysis services to Landing markets so I'm also part of the human factor so this is just to say that you know humans are not bad I definitely believe that at this point the optimal way is to get Services you know from you know living individuals not only from software but in the protocol it's definitely something you know we aim to optimize automate and eventually eliminate as much as possible the human factor of everything and replace people including ourselves with some kind of formula so more background on myself I was CTO of kyber network kyber network was one of the first on-chain swap Services which gave you know very similar user experience as uni swap but instead of a four so instead of a formula we had a very big team of Clans and Traders where we actually quoted prices did a lot of on-chain Market making algorithmic Market making not with a formula and you know only to eventually replace everything also with a formula there and and there it was you know a bit of similar considerations in the sense that our Market making was you know more efficient to Mark some extent but eventually you know replacing humans with software has a lot of also upside and kind of Promise into it uh so some I don't know one of one calls on you know what is even a economic risk management in D5 so this is this is just the basic there's a you know huge layer of smart contact risk and auditing uh which is more elastic security but in the context of Landing markets so risk management is a lot about keeping the landing Market solvent making sure that users will be able to repay their loans so it's a lot about setting the loan to value parameter how much you can borrow against certain collateral when you will get liquidated what will be the interest rate this of course you know has both quantitative considerations such as the market conditions also qualitative considerations such as you know the safety of the price Oracle how decentralized the asset is whether it is controlled you know by some multisigo even eoa or whether it is controlled by some Central decentralized governance Etc and you know so this is a pretty maybe scary diagram but still very high level one so it's mostly to show that almost that it in every step of the way you have so-called human intervention which is again perfectly fine because you can argue you know like as the other community has the right to do what other token holders want to do a further they might you know know or familiar with the other ecosystem and Community much better than you know small smart contract can ever also right in principle it's definitely the right to kind of balance the risk reward you know like eventually the more risky you you will become potentially your Revenue will increase and this is definitely something you know a person with you know free we won't should be able to decide over the outcomes so it might be very kind of optimal to have a as a community as a Dao but eventually is a user you get something that is much more you know so it's not centralized because many people control it but it very resembles to how trade fire is working and basically you get something like on-chain Finance rather than defy so you just replace uh you know all the not even evil actors but all the political actors with the new political actors you do have a bit of more transparency but again not much at the end of the day and sell when you know like in times of Crisis if you let human you know make the decision then it might be right it might be wrong it's in a way less fair and you know less transparent as and as a user you don't really you know you you don't have the guarantee of how things will be executed as opposed to you know uni swap user or user in other smart contracts uh so yeah so the question is if can we replace humans with a formula so here you know also coming from the risk now and you know we are in the trenches we do give paid services to platforms we don't think that you know kind of in the near future automated service or smart contact or formula will be able to outperform you know Human Services a so it will definitely have to come on the expense on the expense of precision potentially even additional risk but again if you take uniswap as a the analogy so we do think it has an added value and you know we do think that eventually part of the default Revolution is that and you know without that it won't be really difficult it would be some kind of Unchained Finance which by domain by the way might also have you know like a home you know like a right of its own it might even be very big but if you really want to follow you know like the defy ethos we want devs to be or eventually piece of code uh to do the execution give guarantees to the user makes things more predictable then you know we think this is the this will eventually be one of the dominating approaches uh so the thing right now there are still a lot of missing you know building blocks to do it and of course you know research and confidence are one major issue there but we do believe you know like the the room of Risk Service vendors is going there are a lot of smart people in the field so it should be kind of feasible to kind of approximate and find you know get into a consensus around a standard formulas for risk and developing you know completely open source a smart contracts that could be used by anyone another missing piece is the actual data okay so right now learning markets rely only on spot price data from changing or other vendors a lot of risk related data is currently not on chain for example Dex liquidity but also price volatility other aspects and this is our Focus site now on the protocol this kind of identifying what are the important input inputs for a risk analysis and to put them on chain in a decentralized manner the first use case which will be launched and it will be highly experimental also a highly centralized will be some kind of LTV recommendation engine well again at a very high level it will just will be able to support and to give a collateral Factor recommendations uh into an asset based on some on-chain data that will upload and maintain again this this will initially be very experimental also very simplified so initially it will we will recommend people to use it only as kind of an indicator and not actually you know like build smart contacts around it but the vision is totally for the long run to be able to automate all these things to be able to replace a lot of the Dow Decisions by automated smart contact decisions and yeah and so this will be a kind of a gradual process you know confidence will have to be built and initially you know initially human decisions will have to be taking uh very frequently but but this kind of presentation will make things you know much more transparent and can be can give rise to semi-automated process and then eventually to a fully automated one uh so again high level architecture here could be the relevant data feeds are put on the blockchain such as liquidity and volatility the Dao or someone you know does set some pre-approved template for calculating the tvl with some formula also potentially putting a some risk you know some risk appetite Factor there and then the ltvs on the landing Market are adjusted automatically or maybe semi-automatically with some vital rights again even in this initial phase the Dow will still have full control but at least here it will be very obvious and very observable whenever the Dow actually deviates from the standard policy a long-term use case you know we can Envision a future where even listing decisions in a learning Market or taking automatically or semi-automatically by a smart contract where the Dow only has to decide on some minimum requirements such as historical liquidity historical volatility and then get listed of course for the sustainable future you know certain aspects such as you know whether the new assets smart contact is safe or not whether the asset team will you know rockpool so you know we are still long way from even knowing how to make these decisions you know kind of more automated and objectives but this this is definitely part of the vision but again starting with some quantitative metrics is the way to start the in our our opinion uh so some comparison you know like between the two approaches so right now the process is very political and again politics also have upsides not only downsides so it does goes both way but very much against the defy ethos uh the process right now is is also very manual which could also you know boss can give rise to some mistakes and actually we saw some configuration errors actually both on average and I think moonwell or something that actually results uh you know just you know missed few decimals and as a result few tens of thousands of dollars got liquidated just because of you know human errors but Manuel also typically means very governance survey right now the kind of standard is that every such change has to go through a governance vote governance vote typically as a time lock the vote itself might take you know few days you need a quorum for every such vote so this is very heavy and also also prevents a real-time responses and this is something we saw is so we saw it on average they are to liquidate a lot of CRV uh and you know the dial cannot respond in less than five days time I think there okay so so it's both very tedious and both prevent the ability to respond in real time so so I after that you know so I've had some committee now with some multi-stick that can respond quicker but but then you know it give rise to the fact that okay now it's maybe more OPEC because okay sure we have a multisig and so you have form discussions on you know what to do and when but eventually you know it's all about humans and you know who might behave differently because you know not because they are bad but because they think it makes more sense to kind of deviate from certain policy which again might be more optimal but not very difficult and you know decrease the predictability of everything automation uh so you know we have a more neutral risk assessment it could be better could be worse but it's neutral and fair it's automated so you can respond in a timely manner and you know as all the defy uh properties that we are all you know eventually here for self-execution transparency decentralization so yeah if you want to learn more so check that out check us out at the depot.org kind of concentrate a lot of the actual research the protocol is more the decentralization layer I'm in Taipei until Wednesday so also reach out to me either on telegram or Twitter and yeah with that I'm concluding and if there are any questions [Applause] so the first use case you gave was working with lending platforms to set the long-term loan to value uh could this be adapted to set the LTV for a decentralized derivative protocols yeah so so right now most of our research is focused on landing and stable coins yeah but definitely is a vision yeah that definitely also for derivatives also in the risk now you know like as a service we also analyze derivative in terms of the protocol and making it automated so it's definitely part of the vision right now we are mostly focused on lending though but yeah definitely also the images thank you any other questions if no uh let's uh plus again thank you thank you Karen and [Music] [Applause]