The Dark Side of Perp DEXs: Rebuilding DeFi’s Market Layer | Douglas Colkitt - Ambient Finance

All right, we have our next speaker with us, Douglas Kolkit, uh, founding contributor of Fogo and the founder of Ambient Finance. He's going to be talking about how we're going to fix per dexes so that 1010 doesn't ever happen again, basically, right? We're going to fix the the flash crashes. No more ADL cascades and more exciting things that will make our lives safer and better in DeFi. Thank you, Douglas.

Thank you.

Here you go.

Hi, thanks uh thanks everyone for coming out today. Uh so yeah, my name is Doug Kolkat. I'm uh the founder of Ambient Finance uh and founding contributor of Fogo, which is a a fast uh fast new L1. and ambient finance. Uh we built basically a large AMM spot uh decks and now we're in the perp decks space.

Actually, you guys know why the uh Okay, awesome. So, I guess my talk today is called the dark side of perplexus, but before I get into that, I want to really say the light side of perplexus, which are like un arguably perplexus are the most successful DeFi product in history, right? Like we've actually onboarded tons of people. They're real use cases for the chain. People are paying real money for financial application here.

So especially hyperlquid is these are amazingly successful applications and come really far and I don't don't want to negate that part where we get into the dark side of perplexus and say first question we ask is why has there been no perp so we think back to like 2020 defi summer we had these new kind of primitives curve unis swap uh compound synthetics and right there was cool thing about that is they all built upon each other and one success led to the success of other products, but we haven't really seen that in perp space yet. We've seen individual applications and protocols and products get successful, but that doesn't seem to be spilling over. And so this chart is, for example, uh protocols on hyperlquid fee revenue. You can see it's pretty much all fee revenue is still just going to hyper core itself. Like there's this hypervm ecosystem.

It's been out more than a year and it hasn't really seen kind of anything close to the explosion we saw with DeFi summer, right? So it's kind of this weird spot where crypto's achieved very DeFi has achieved a very successful product but we don't have anything right and the crypto market's still kind of dead you know besides this one vertical so kind of what happens today in uh perex world is total vertical integration right so uh you know you go to the tradfi world there's this concept of clearing house and exchanges and you might trade uh you trade a derivative at one exchange but there's a centralized clearing house and you can move positions between them. uh right now in in crypto even even in decentralized systems these one protocol handles everything matching uh execution margin liquidation ADL even usually UI right um so there's single execution path there's no venuetovenue uh competition per positions aren't portable and you really can't do anything with them the same way you can do with uh ERC20 tokens so we call this is kind of composability ility uh costs play like we do have some things that are hooking into perp dexus right especially hyperlquid some on lighter now but these are really APIs they're not smart contracts right even hyperevm and hyper core like the hyper core being the actual perex and hypervm being where you're supposed to build on top of it hypervm is a secondass citizen actually trying to talk between the two is you know anyone who's been a developer knows it's very tough uh you can't reuse positions so I can't take positions I can't rehypothecate them. I can't slice and tunch risk. So everything gets stuck in a single stack.

You don't have these DeFi DeFi Legos, right? That really are the core of why DeFi was successful to begin with. Uh and there's a hidden tax to this. This is uh 1010 ADL. I don't think I have to tell anyone about like how bad that was, but one thing finance is supposed to do in general in DeFi especially is trunch risk, slice risk, and repackage risk where it's supposed to be and who's most likely to hold the risk.

So ADL is one aspect of this and when it's not composable uh what that means is you have people with different levels of risk exposure. What happened on 1010 was a lot of people had exposure uh who really couldn't handle ADL, right? You have market makers, you have people running delta neutral strategies, they get unhedged on one venue and then they're really at risk at another and uh you have these centralized systems that are trying to rank people. There's really no way for people for example there's one dimension they can't trade that. Uh same thing with liquidations, right?

like because there's no way to go out and get additional leverage from riskier protocols or trunch risk, trunch clearing risk, uh you basically lock everyone into a system that's too DeFi or 2D gen for uh kind of the more conservative professional players but too risky for uh the more institutional players. So one size doesn't fit all uh here and uh right when it blows up it it tends to blow up bad. So kind of the way we've been thinking about this is you really need to split the system into two parts. Uh one is uh the clearing house level and the other is the execution level. So you have you have a clearing system you have clearing house and that handles things you think about anytime you hold a position.

So you want your margin funding liquidations ADL all of that stuff. And that part I think makes the most sense to be fully on chain. What that unlocks is you can start doing stuff with your positions. You now have the equivalent of an ERC20 token for your per positions and including right you can lock those into additional funding systems, additional leverage systems. You can take uh you get leverage outside the core clearing house because it's composable.

Um and that works fine right because clearing and settlement is actually not a very high frequency system right like a lot of trady systems they only settle at the end of the day. That's suboptimal. we like faster in crypto, but you really don't need a super high frequency system. The other side of this is execution, right? Execution is kind of a different game than clearing.

Uh it's probably always going to be the case that somewhat off-chain solutions or quasi off-chain solutions have advantage here, right? People are going to want to execute on very very fast systems that are just fundamentally hard for blockchains to deliver on. Um and maybe not all the time, but sometimes, right? So if you don't have kind of a pathway for that off-chain execution uh you know people go back to kind of siloed centralized vertically integrated systems. So we think makes the most sense is have clearing systems that are extremely modular and to allow competition on the execution space between different different venues.

Um so in terms of this who owns what right when when you go into this kind of onchain clearing house system right so uh again I think it makes sense keep everything onchain as much as possible and uh the other part of that is right that gives you the part you're really worried about from I'll say centralized attack is the clearing house side I don't necessarily care as much if my limit orders are I'm trusting somebody to give me good execution I care a little bit but what I really care about is okay you can't liquidate me if my position's uh not underwater you can't uh you can't give me a different position you can't say I don't have margin when I do have margin you can't seize my margin and right these are all the things that we want very onchain systems in in fully generalized chains not kind of trust me bro uh perex uh you know perpex application chains and again this is this is possible you can have two systems kind of bridge bridge this gap so what do you get what do you get from like open execution like when you have something like this. I think like the most successful system has been been in spot world. Uh so Salana spot execution has probably been a great example of this right and it doesn't work for per right now but right kind of gives us gives us a layout of what something looks like right and you have different different layers to the system. So right like you have aggregators and they compete for the end user and who what aggregators do users trust most and those you trust those aggregators to route you to the best venue. um right you have execution venues themselves right and there's a diversity here and as as pers expand and they probably will keep expanding into things outside of traditional crypto assets some of those assets have very different financial properties right like a two-year treasury does not trade the same as uh you know a super leveraged shitcoin so the market structures that make sense for those respective things uh might be different right so you might see offchain cloves you might see onchain uh what we call dual flow batch auctions which basically a blockchain friendly version of a CLAB.

Uh prop amms, right? We just had a great talk about them. They don't work in per at all because the problem with a prop amm is uh it isn't the clearing house, right? And if it was the clearing house, it wouldn't work at all because no one can come in and provide liquidity. So you could be stuck in positions if they turned off, right?

So if you want something like a prop amm for pers, which makes a lot of sense for especially longer tail assets, uh you need something like a centralized clearing house. Um and then right like stuff like GMX like that's onchain virtual AMMs that have use user supported vaults those are actually great assets right because not everyone can provide liquidity in a book but an AMM anyone can deposit to a vault and you know even if execution isn't as good for longtail assets a lot of times liquidity subsidized and we've seen that in spot world uh you uh you just can't do it right in in that kind of world. So once you have this once you have a clearing house layer that stays concerned with its aspects when it plugs in right you can take all the innovations we've seen in onchain spot trading push them push them in the per world right so is what what basically do you get from this right you get spotlight competition the amazing thing about spot is I can buy at unis swap I can sell at 0x the next day I'm not locked into one who I execute with right now is uh it's a date I'm not marrying that persons is different. I open a position at lighter. I can't go and sell that at Pacifica, right?

So per right now create a system where you get kind of virtual monopolies. Um and you know that's not great. So once you have this, right, you open up competition. The other thing is you get portable positions. So you can not only take positions, you can start slicing and tunching the risk associated with that position.

What are risks associated with position? liquidation ADL uh you know whether you have to post margin right so you get different players who have different sensitivities to those things you can start tokenizing those positions and you can start uh writing say ADL insurance on top of it you can go to another protocol and say will you compensate me you know if I get a certain amount of ADL will you guarantee my position stays open uh will you give me higher leverage on top of what the clearing house does and right you can isolate kind of that longtail of risk uh then you these once you have these right you start getting real Legos right and this is what this is what we want from DeFi we want Legos we want things that go together and this is how crypto wins because one product takes off you know it's not just that product and only that product it's a bunch of things on top of it right and and that kind of open innovation is really the only way like we get we get to the point uh where per actually start feeling like a decentralized system and then yeah finally you just faster innovation better execution better risk management that's kind kind of that's kind of uh our vision for pers and I think that's how you how you fix the perpex landscape and kind of the dark side issues there. So I have about minute left happy uh happy to take any questions.

Do we have the questions? Do we have a microphone? There's a question. Yay. Applause for the question.

Hey.

Hey. Whoa. [laughter] Hey. Uh, the vision was really compelling, but uh in practice, how does one actually make it happen? Like how do you align all the incentives?

Yeah.

Yeah, that's a that's a great question. H so how it actually looks uh is that you move kind of the clearing house layer and so you have to do all of this stuff fully on chain. And that's that's definitely not trivial. Um but that can be done right and at you know faster chains uh Salana like Salana like speeds and Salana like costs you can do that so it's not feasible on mainet probably but it is feasible on fast cheap chains uh the settlement part the execution part is really hard but I I think the key thing is right you need good primitives to allow uh kind of those execution systems to hook into the clearing systems there's a lot of complexity around like making sure you're when you're executing people they have the margin there so you have to use intent systems you have to lock collateral you We we've been pretty deep out work on it. I could probably go to 10 times as length of this talk, but uh yeah, it's a great question.

It's a lot of easy stuff to say and definitely hard. Uh you have to really think about these things in practice.

Thank you so much, Doug.

Awesome. Thank you.

All the deep research and thinking. Applause for Doug.