Panel on DeFi and Risks on Leverage moderated by Sarang Parikh

Transcript

yep um hey guys how's it going so everyone like we're going to talk about uh leverage right it's a wonderful presentation so so much uh it will carry on the floor for the audience as well right so let's start with like you know a small introduction and uh what do you like do like what does your product do and uh what do you think is uh leverage and composability is it like a uh curse disguises a cure for defi you know because what other the hacks we have seen and everything like what is your opinion so I'm going to start I am saring I used to lead engineering at Sushi swap uh now I'm an adviser I help out other projects and for me uh it's always been a cure it's never been a c hi everyone I'm Eugene I work at

Ellipsis Labs we are contributors to Phoenix which is a limit order Book Spot deck uh no leverage on salana currently doing few hundred million dollar of volume per day I find leverage very scary it's very difficult to get right uh but it's clearly something that's demanded by the market hey folks I'm mod Gupta um I lead security at polygon by day and D leverage Trader by night uh Leverage is basically like the code comes to mind with great power comes great responsibility it can wreck you but it can also be very fun if use correctly yes I already have a speech so my name is Mikel I am original inventor of gearbox protocol we making this

composable leverage and from my point leverage what is very very significant brings Capital efficiency to the industry and efficient market always a good public value for everyone because more efficiency means more money more safety and so on yeah um so that brings to my next question with uh efficiency it comes like Capital at use right and there is this Leverage de cycle that people keep leveraging like what do you see with Pendle right like you have multiple folds of actual leverage to get that point so like when Pendle actually started like it's when it had the first pmf was when it takes rocket pit it deposits into balancer then balancer

deposits into AA and Aura then deposit into Pendle to get that you know all that leverage yield right and that could really turn into his disasters anyone of the you know like building blocks fail so when do you think is like a stopping point or you should analyze okay like is this Leverage is worth it and how do you analyze it because now there are multiple fronts of the attack right I think Leverage is always risky and should be done by only professional players and it's how maybe current markets exist everywhere because basically crypto is not mature at all and people who have some experience or knowledge could go through the leverage

and of course we should build this risk Frameworks to make it safer to better do all risk management and provide all needed information for decision making on the other side just imagine with leverage you can really help pretty good protocols grow faster and deliver value faster so you are great to talk about all this inter connectivity it could be a systematic risk it why we really need researchers and scientists to build a robust Frameworks to compete Ute and do it on the mass approval side but what I really like in a crypto that everything is transparent you should not trust like any uh centralized exchange and there reports you can do your own research

through the other scan it's it's amazing so do not trust Leverage is risky but you can check everything on chain and make your decision with the whole information in comparison with all other Financial system so more direct to the question like with the Pendle and most protocols as well like it's leverage is all about risk reward ratio you get more reward but there is more risk and the risk at least in my opinion and even in the Market's opinion wasn't worth it till points came in and once the points came in the reward just shot up while the risk remained same and you saw a boom in usage for all of these protocols which

were providing leverage so yeah leverage can be good good when the risk reward ratio is good it it's super bad when it's bad and it's up to the market to decide yeah there's like a bunch of different forms of risk and I think one of them that you know we think about from the user perspective is I put my money into some leverage protocol and How likely is it that I get liquidated and then there's other form of more systemic risk where supposing everyone has you know their their liquidation points all the same place like everyone exactly 10x levered on some uh cursed like reaking eth 10 times over um and everyone gets liquidated at the same time what happens to the whole system like do the protocols themselves go insolvent and this is uh you know going back to your point on transparency if

everything is actually fully on chain uh it is possible to Monitor and detect these types of things uh there are still are a couple problems here where like one there's not too much of an incentive for someone to go out and do this it's kind of this like think this work uh there are some folks who who do it um for free like the the paret guys for example always have uh these these great Dash boards around liquidation levels on on a and other leverage protocols but I think the The Leverage builds up faster than dashboards can can build up and um there's also this incentive misalignment between the developers and the users where yeah if I'm just starting out building a new protocol uh I may be incentivized to give my users much more

leverage than what is going to be safe for the the the system because I'm trying to juice the TBL and gain some attention and that problem worries me me a lot more because I don't really see any tractable ways to to fix it I think that's getting worse because now this new protocols have points which gets levered I tried to pay people in points but they just don't accept it I don't know why but they keep on going on pendal and farming so yeah uh but yeah that that brings me to the question right like so you mentioned the factors that could uh you know have systematic risk based on your protocol right and how your protocol functions but there could be more factors which probably might not be your control let's say uh what happened to I think like hyper liquid like few days back like that

Oracle issue where they had to you know they some of the position got liquidated or you know there uh issues at some chains where they stop for some time and they have to resume it back uh and that for leverage protocols it's it's it's almost very bad right so uh how do you think like people should build leverage protocols in a way to be sustained because you can't guarantee in almost up time for almost all the new chains right so maybe as an app Dev you we have to be careful about how do we build it so if you guys have you know have some thoughts around it I mean I'm an application developer but our product today is 100% spot so we don't have to worry about any of these leverage issue so I'm kind of um you know shouting from the bleachers here

but there's a bunch of things on the on the infra side and really just any dependencies that your protocol has whether they're on other assets you sort of care about their return profiles you care about uh how safe the Oracle that you're using is and you need to be able to model all that type type of risk you care about things like the chain going down uh and the like it's pretty easy to model these things like independently and reason about them one at a time but actually they interact with each other in these very weird ways where sort of all of these tail events tend to happen at the same time so for example if uh some blockchain that has a leverage protocol the blockchain itself goes down some other things you might expect to happen are like oh the Oracle update is going

to be stale the Oracle update the next one is probably more likely to be inaccurate um um the volatility that you expect during the time when uh the chain is down you should expect to be significantly higher as well and these things are all just super super correlated and if the goal is to design a protocol leverage protocol that's going to be robust to all these things you need to be robust to all of them breaking like exactly at the same time you need to be robust to other people coming in and using your product in a composable way to like maximize the risk involved or like maximize the degeneracy involved and ideally you can be robust to that so please you you go ahead tell tell us how Gbox do it yeah I think it's a

pretty interesting stuff related all this uh risk events I definitely believe and I know that some teams all work on different simulations but Market simulations it's a very very complex stuff because you never know how it happens and of course I think this pure communication of course if you go to a pretty eurolab or l2s you should assume that there are more infrastructure risk as you said that L2 could be blocked for some period of time or do not respond and so on and here I believe another interesting part what we already told is how we can throttle this L2 when it

comes back because it's a very very interesting problem you have a state and this state is Frozen for many many hours here and then L2 comes back to life and in comparison with mainnet for example it's a huge price difference if you allow users to make one trade for the arbitrageur to take the all profits to sync the levels between mainnet and L2 it means that these guys get unfair profit and if you have a sequencer maybe this profit could be shared with sequencer but all other users could get losses and here probably maybe on the

protocol level or sequenc level we could find the solution how to throttle these networks to get a equal access to the capital when it will synchronize the prices between Frozen l2s and real market and it's like a technical problem but probably if we could build this in a more robust way it could reduce the losses and do not make like a huge profits in one Arbitrage your hands yeah I I think like most sequences are centralized as of now right so this can be done uh with much effort you know yeah but like at the same time there's more effort being put on decentralizing the sequences so this would be counter productive in that sense um but like

from a if I was veloping a leverage protocol like I would build something like a heartbeat so that means like the protocol would kind of just stop uh trading if there is no heartbeat from uh Oracle SL liquidation agents and you can then have still time based price feeds like whenever the chain comes back the Oracle would still uh feed in the prices with relevant times and the Liquidator would liquidate at those price points um so it does add another component and and potentially a bit of centralization risk but if you can make it like a decentralized network sort of uh with like for example how chain link agents are supposed to work um then it can work

out and I wonder like how many of the protocols actually do that uh so it's it's also a question to ask when we actually use these protocols and there there's this real risk right and if these protocols do that if not then you know they should adopt to try to do something like this you know um so yeah and there is uh other kind of risks right like now uh when we've talked about like oracles right so all of these leverage protocols want to have as many assets as they can right but that comes with a cost right and not even just small one but big ones as well which take up like a very high position such as what what happened with a and curve Saga right where a had a lot of curve exposure right and in that case like

what can be the ways where you can avoid getting into this situation or how could you as a user be informed on okay like this this token doesn't seem that liquid right and even when the TAA thing happened like the liquidity could pulled off the curve right sometimes you might not even know so uh how do you get this with the shared risk versus the isolated risk in leverage y so I mean the reality is 99.99% of the users don't care SL don't know how to even care about this uh and that's why we really need um the protocols to have like a security Council or leverage Council or governance or whatever you want to call it to handle these parameters U and it's

the standard in this industry like uh vendors like Gauntlet chaos Labs chain risk all of these are like it's their job to um like help set the parameters correctly so protocols can't like don't go in bad debt risk like this um obviously like Gauntlet didn't didn't manage to pull off the perfect job with our so we got in the situation um but like it it's if they can't do it we can't expect the user to be able to do it so we just have to have professional people actually managing risk there yeah I really like the name uh leverage uh you know like a team like or like Leverage Council we can just drop the Lev and just call it rage Council I think that'll be better uh but yeah yeah

from my point I can't agree that any leverage protocol want to maximize the quantity of assets I believe each leverage protocol wants totally the opposite at least in gearbox my point was always we want to pick the best ones to generate high yield in a safe way because we shouldn't have 1,000 different assets and really have high risk we want to pick what really users want and then we kick them out all other stuff because basically if no one go to positions there is no reason to keep it in your protocol you can simply delete or set the lower limit and so on and

what I already told with a quar system we can really make it very very flexible how much collateral we can use for any particular asset it's a good point but here I definitely believe that users could vote what they need the leverage protocol should provide what they need and the key metric for any leverage protocol is utilization it means that you really give required asset with correct limit and I definitely agree that professional risk players like C slabs we are starting to work with them I think in a month or so could really focus on all these parameters to make it

safer but please do not add as many as you can it always under control and consequences could be pretty bad for anyone who want to add whole Market to one protocol yeah I really like the giving the power to the people like you know for voting my next question is going to be based on that but if you want to add in something you know uh I really don't have too much to add I think those are both good answers uh just like to reiterate that it's pretty unrealistic to expect the users to do all the risk management here and there has to be something like done at the protocol layer you can't just blame the users when the protocol blows up yeah so my my next question is okay like we have governance and tokens and votings right so

uh this was also tailored towards the Athena people like uh um like um the the question was like okay like if there is v aena in existence for instance right and Athena is super big right uh in a utopian world like like 5 years from now Athena is most one of the most dominant stable coin there is VE Athena now uh who would be the uh and atha lets you the vtina lets you control everything about the protocol or most of the crucial parameters crucial par parameters as which exchange should we have a lot of liquidity on which exchange should we open shorts right and then if binance comes or like some other exchange comes and buys off a lot of vatina and use that against some other exchange or use that to cause issue for

the protocol how would you do that so like is is it a good idea to give governance uh you know like uh would like use uh set critical parameters using the governance and voting I mean I think there should be limits on governance no matter what TB if uh token based governance makes sense for anything at all but it certainly shouldn't have all of the Power I think the the obvious example here is uh if the token voters have complete power over everything that the protocol does you can always do the thing where 51% collude to steal money from the remaining 49% or whatever your voting threshold is set to in this particular case I mean I would imagine some like

pretty robust risk management system 5 years into the future where you are deploying your Capital across different centralized and decentralized exchanges based on some combination of available liquidity and the um the current funding rates on all these different exchanges with some assumptions on like how long it's going to take to to move the capital from one place to another if you need to do it if that was delegated to governance then I think there's a lot of attacks that are available not really anything too satisfying on the protocol itself there's always weird things that happen when the protocol sort of like outgrows the Bas layer so here one example would be like if the if Athena represents like a very very large portion of the open interest on a particular exchange there

can be like potentially very bad interactions that happen there either bad for Athena or bad for the centralized exchange or both and so that's type of thing where if I was a very very cynical large exchange uh with the ability to buy up the voting power might want to consider doing something like attacking other centralized exchanges uh using this so yeah in short your answer is like they should be the governance should not have access to critical uh protocol Parts yeah I think uh government should not be touching stuff like this yeah I mean token governance hasn't really worked too well for any protocol there are lots of um like the reality is in this space there are maybe one% of

the actual teams were doing token governance where it's actually decentralized everywhere else you go it's like you look at the tokenomics foundation own 66% of the tokens or whatever and then it's token voting let's go or one of the VC's owns majority share so uh there are very few players who do it right and it's very hard to do it right um I think ideally what should happen is like it should be more uh transparent of the reality so I really like to compare it with actual democracy and I think like we have figured out how democracy should work more or less over time and we can just reuse that like all of the token holders or whatever come together vote once every few years or whatever and select

one like group of people that's going to lead and then they lead the protocol like the actual people all the power they have is to change the people who are leading not make individual day-to-day decisions they don't have the context for that um and I think that model works very well the team who got selected has some Independence uh can work and can take tough decisions and so on because lots of these decisions are security related or things which you cannot share publicly uh for voting and so on so I think it's it's very important to have like one like closed Council but at the end of the day the token holders should have some say so I think the balance is where they are allowed to select the council and if

they don't like how the council is working they can change it yeah so instead of like per word governance should go towards representative form of you know governance yeah basically I'm also not so good in Risk I'm focused on more technical side and Engineering but uh I definitely agree that the protocols should be managed by token holders and gearbox is Dow manage protocol from day Zero because even it was deployed people voted what we should add and what not and of course we should attract a pretty strict strong players like gaunlet or cow slabs in our case which could really analyze and provide some uh suggestions and proposals for Z

da but interesting F related to enena that enena was added to the moro uh through the maker and maker asked to hard quot as I know correctly to make an usde and die Oracle one to one so basically in Moro We have a market with one b million die on one side and one around 1 million USD on the other side when Oracle is hardcoded which also give some ideas that maybe it's a good sign I don't know exactly here but makeer trust the Cena for such so huge amount of money which also make

it's trusted from maybe our side but it's a pretty hard to make all the simulation and risk as I said I'm more Eng genical part and here of course professional should analy pleas yeah that's good point so yeah I think like uh almost come down to the end of the panel does anybody has any questions I guess not so I think that's it uh thank thanks for coming down and you know uh giving us insiteful thoughts thank you thank you thank you s