Perpetual Swaps 101 - with Su Zhu and Hasu

Transcript

Welcome to Uncommon Core, where we explore the big ideas in crypto from first principles. For today's episode, I sat down with Su Zhu to learn all about perpetual swaps. There's probably not a person in this space who has more authority on this topic than him. The first half of the episode serves as a general introduction to perpetual swaps, while in the second half we explore how perpetual swaps are coming to DeFi and why they are the single biggest primitive that is still missing from the DeFi puzzle.

While the episode can get pretty technical at times, we try to follow a logical build-up, so no prior knowledge should be required. Please note that I put some links to definitions and follow-up reading into the show notes. Enjoy the episode. Uh to start out the discussion, Su, what is a perpetual swap and how does it work? Uh perpetual swap is the idea that the buyer is buying a synthetic exposure to an asset and the seller is selling or shorting exposure to that same asset. So, the the word perpetual there has a important meaning, which is that it has

no expiration. Unlike a futures contract where it'll have a predetermined date at which the derivative no longer exists and gets settled, a perpetual swap seeks to maintain the exposure for the buyer and seller perpetually. And and so that that is in a gist what a perpetual swap is. There's a lot of different ways that um it can be traded and there's a lot of different ways that it's structured, but uh fundamentally, it's a way for uh participants in a market to come together and exchange risk.

Yeah, so maybe to slightly rephrase what what you said. So, for regular futures, the um the price of a contract tends to gradually converge with the the spot price or the index price as expiry comes closer, right? Um but this is not the case in a perpetual swap because it never settles. So, um can you explain maybe how how it is done in perpetual swaps? How the um how the perpetual contract can successfully track the underlying index price?

Sure. So, the arbitrage mechanism of a future is, as you alluded to, well well defined, right? Because the contract will essentially define the index at which the derivative will necessarily settle on its expiry date. So, if it's a cash settled future, um that settles on, let's say, the TWAP of four spot exchanges, then on that day, everyone knows that that will be the price that these derivatives end at. So, from there, people can buy the spot, sell the future, and vice versa to to replicate the risk

that they're getting. With a perpetual swap, because there's no expiry, there has to be some other incentive for buyers and sellers to meet the natural demand of the market. So, if the market generally wants to be long, then you expect perpetual swaps to need to incentivize sellers to come in in order for that market to uh happen. So, there's a few ways that perpetual swaps can do this. Uh the most popular way in cryptocurrency is to have a scheduled funding payment between buyers and sellers.

So, in that example, if people are bullish, then swaps will be generally paying positive funding payments, whether every 8 hours, such as in the case of BitMEX, OKEx, and Huobi, or every hour in the case of FTX and Kraken Futures, and then continuously in the case of Deribit. But, just in general, the idea that the buyers are going to have to pay the sellers some amount of money every period in order for the seller to stay in that position. So, so that creates a arbitrage opportunity for the seller, right? Cuz the seller can come in and short these perpetuals, buy the spot, and now they have essentially a crypto dollar, right? They've they've

regenerated the exposure of $1, and they're collecting interests while doing so. And if they collect an interest rate that's attractive to them, then they're happy to be in that position. And and if they're not, then they'll unwind it. So, that that in a nutshell is kind of how perpetual swaps became very popular uh because you could uh you could buy or sell a perpetual swap and know that your exposure never disappears. Mhm. Right. So, you think you're talking, for example, about the case where I deposit uh BTC on a derivatives exchange, so I'm long my collateral, and then I sell the perpetual swap contract for the same position size

as my collateral, and thereby neutralize my long exposure and create um a crypto dollar correct? Yeah exactly. Okay. Okay. So, for the funding rates, so a mental model that I that has worked well for me was to not actually see these as perpetual at all, but just um imagine that So, let's say in the example where you have a funding rate every 8 hours to just see this as distinct futures contracts um, that settle every 8 hours and then you're automatically rolled over into into a new future. Do you think that's a sound model?

It's not quite It's not quite the model I would use. I kind of I get where you're trying to get at, but I think the main difference is that in the case of a cash settled future, you know that the swap will converge to the index price. Whereas in the case of a perpetual, you don't know that the raw price will converge to the index. In In fact, when the market is very bullish or very bearish, that price will not converge, but you'll get paid something in lieu of conversions. So, in these formulas, they've been generally made quite aggressively to pay arbitrageurs because they want to protect the integrity of their market and the trust that people have that it will track the index.

Mhm. That it will track the underlying. So, these payments are large enough that they seem like they've essentially paid for the difference between the swap price the the observed swap price, and the observed index price. So, I I think you're right, but I think there's some nuance there. Where it doesn't have to converge. Got you. Got you. So, there's this um, persistent myth um, that I also I also subscribed to for some time before like learning uh, what's actually true, that um, perpetual swap contracts were invented in the cryptocurrency space, but but you

actually taught me that it's not true. So, can maybe give us a quick overview about the the history of the perpetual swap contract. Sure. So, I mean, it generally depends on how specific we want to be on what is a professional swap, right? Some people might say, "Well, professional swap in its modern sense was invented in crypto." Some people might say, "Well, it has a predecessor or it has a ancestor which is the idea of an index swap or equity-linked swap. Basically, the idea that two counterparties meet in the market and decide to sign a paperwork to exchange cash flows,

Mhm. right? That That That's pretty much all a professional swap is. So, if you take that sort of philosophical view of what a swap is, then obviously they've been around for probably 30, 40, 50 years, right? Because since the beginning of the modern financial markets, there's been a lot of demand for sophisticated counterparties to make their risk exposures synthetic rather than actual. And Mhm.

these reasons range from tax uh efficiency to capital efficiency to more broad offshoring of risk. There's a whole host of reasons why participants wanted this in the OTC market. But, the reality became that the the vast majority of, for instance, equity exposure of European funds into Asian equities is not via owning Asian stocks, it's via owning equity swaps. So, these swaps will be collecting and receiving payments based on some criteria, and the buyer gets exposure to the index, the seller gets exposure to the negative index. So, the

sellers are typically banks, right? Investment banks, they'll sell these swaps, and then they'll buy the underlying markets. Mhm. Okay. And the most common benchmark is the MSCI index, the Morgan Stanley composite index. So, Yeah. you know, there's trillions of dollars in these indices and these swaps are essentially perpetual swaps. Uh because there's a funding payment. Uh It's benchmarked to LIBOR plus minus something and there's also uh an index. And so, Right.

this concept is not new. This concept is not new, right? The main reason participants like this is because there's no need to roll it, right? There's simply a payment. Uh So, the problem with the problem with futures is that participants need to constantly re-meet in the market and roll their position. And that that creates a lot of uncertainty because what if that market shifts very quickly? What if that roll is not available? And so, as a result, you you tended to see uh in the retail market as well, uh a key predecessor would be the contract

for difference or the CFD. And this is a very popular market in the UK and Australia, pretty much in markets where CFDs also get a preferential tax treatment because they're considered bets and gambling is sort of lionized in the Anglo in the Anglosphere. Basically, it's not taxed, right? come from that. Yeah, so, you know, the idea being that if you're if you're a sports better, shouldn't be taxed on winnings cuz you're negative EV anyways. And Right.

So, but but then as a result, you know, when you invest in stocks, you do get taxed because governments tend to think that when people make money from investing, they should pay some of that back. So, it's kind of an arbitrage because now if you say, "Well, I'm not investing, I'm betting that the number will go up." Then they say, "Okay, well, yeah, then then what do I do?" So, as a result, CFD's very popular in these countries and same concept. There There's an overnight funding rate. Uh you know, one one key difference is that usually these are not peer-to-peer, but these are peer-to-dealer, customer-to-dealer. So, the dealer sets the funding rates. The dealer takes a bit ass spread on that funding rate. And uh that's It's not a It's not a

peer-to-peer market. Uh in the Japanese market, perpetual swaps are also very popular. Uh also known as CFDs in that market because again, there's no expiry. And you could consider the entire FX market to be a perpetual swap market as well because there is a funding payments exchanged every day. Uh They're called the tomorrow-next swaps, tom-next swaps. Basically, the payment that you receive or pay to be long, let's say, dollar versus yen, euro versus USD. These are all payments that are analogous to funding rate payments.

And in the inner If you're in the interbank market, then it is almost like a peer-to-peer market, right? Where there's no bit ass. You can exchange it at the same price. You can buy and sell at the same price. And your exposures are carried forward through time. So, so I think that that being the backdrop, I think the explosion of crypto perpetual swaps uh was made possible by a few things happening at the same time. I think one was that spot markets became close enough in line that you could start constructing indices of them to benchmark a swap against. Because that wasn't always the case, right? There

used to be 5-10% arbitrages between spot exchanges, and it'd be very difficult to construct such an index to that you'd want to benchmark to. I think when BitMEX launched their swap, they had to change their index radically several times, right? Like first they had Bitfinex in it, then they took it out. You know, they had OKCoin and they took it out. And then they, you know, then they made it just Coinbase and they made it just, you know, it was a is a at time is very hard to figure out what that should be. I think also perpetual swaps became uh a strictly better product than what at that time was the weekly futures on OKX.

So before before perpetual swaps everyone traded the weekly OKCoin or OKX futures. And as well as their quarterlies. The problem with the weeklies is that retail clients often forgets to roll their positions. So, if you don't if you don't remember to let's say sell your expiring one week and buy the next one week, then on Friday evening you'll you'll log in and you realize you're not long anymore. You you've already gotten an expire. So, quite often then quite often then the market maker would actually analyze this and then say, "Hey, you know, 30% of open interest is not getting rolled, you know, but all the price action will often be that it, you know, it gets dumped into the expiry print. So, anyone

who doesn't roll their their and if you're long, let's say, you're synthetically saying that you're going to sell your position at any price, right? Because you didn't roll. Yeah, you're forced buyer or forced seller. You're yeah, you're you're basically when you don't roll, it's not that you're doing nothing. You're you're giving the market a a market order. That That's what you're doing. So, you you see basically the market tank into the expiry print and then moon right afterward. Uh this is a very common pattern at that time because market makers realized that this is how you make a ton of money, right? You slam it down. And in theory they should be defending spot, right? Because if you're letting your long expire, you're basically if you're market neutral, you're presuming you're you're saying

the market that you're going to now bid spot to replicate your exposure back to get your your long back, right? So, if you don't defend spot, then the guy who has the opposite position to you, he's going to smash essentially smash the print in his favor against your position, right? He's basically selling your position as low as possible so that you're you're synthetically selling that print. Okay so like a lot of people in crypto talk about the like trading around these expiries of I guess weekly or monthly futures on CME. So is is that basically what you described?

Yeah so it's similar. I mean, the CME print definitely very important. Think think not just psychologically, but also from a flows point of view. Just in terms of if people don't if people don't roll futures right? And sometimes people don't, what they're saying is that they don't want to have it anymore. They they don't want to be long anymore. They don't want to be short anymore. And that that's treated as a as a market order that hits the market. So sometimes this can be very big, sometimes it can be very small. And sometimes people can see this and then anticipate it. So there there's something game theory here. You can't just say because of this then that.

Market's dynamic, right? But there's a lot of important things happening on that print essentially. There's a lot of risk being exchanged or not exchanged based on that print. And just to give an example of how it's hard to interpret sometimes, right? Let's say someone is long a lot of futures on CME and then they seem to be expiring them so that they didn't roll them to the next month. But they could be now buying spot during the TWAP the TWAP window instead as a passive client. So that's actually what some of these guys like Paul Tudor Jones and and some other macro fund managers supposedly did, which is that they wanted to get in fast because they thought Bitcoin was going to moon. So they first bought CME futures because

that's the easiest way to access these. And then later they said, "Well, I want to actually get the Bitcoin spot. How do I get that?" It's a couple of choices, right? One is you could find someone to trade the future versus spot with you because you're already let's say long, I don't know, $200 million worth of Bitcoin through futures, it's going to be annoying to transfer that into spot, right? Well, not actually that annoying because all you have to do is let your futures expire, and then as in the in the expiry window, in the you know, the the expiry definition, you then buy spot Bitcoin every minute to get your exposure back.

And you have no slippage, right? Because the price that you buy spot at is the same as the price that your futures are defined to expire at. So, so in a way, that that mechanism is very powerful. And so, when someone expires their longs, it could mean that they're not they don't want to be long anymore, or it could mean that they're going to switch it into spot, which is also very bullish. So, just to give an example. Got you. So, do you think that the perpetual swap contract is strictly better than monthly or weekly futures? And as a result, is it going to completely crowd out that form of contract?

I think it's not strictly better. I think it's definitely an improvement versus weekly futures for retail. I think it's a very good retail product. I think for institutional players or for more sophisticated players, they also need month-end futures, but they also need quarterly futures. Good example would be to hedge options flows, right? So, options typically concentrate around the month-end and the quarterlies. And if you're an options market maker, you definitely need to trade a future that is in line with your options expiry. Because if you just if you if you just trade perpetual swap against it, you're going to be in big trouble if the

funding chain is a lot, because you have to price in how much swap funding you expect to pay or receive over the interval of your options life, right? And so, so that that process of pricing it in uh you're going to want to see how the actual month-end future were to be traded to price it. So, so that that's one. And then two is that if you're let's say a miner, if you're let's say a a uh spread trader, you you generally want to lock in the duration of your positions in line with your your actual cash flows, right? So

Yeah. let's say you're going to you think you're going to mine this much BTC over 3 months, you probably just want to sell a future and be done with it. You don't want to sell a swap and then have to deal with checking if the funding is in line with what you expect. You know, you kind of just want to do the you kind of just want to do the simplest trade. So there is always going to be a lot of demand for quarterlies. Uh the quarterly open interest on the Chinese exchanges is very high on Huobi and OKEx. It's actually Huobi and some others up is usually higher than BitMEX swap. Mhm. And and the even after introducing swaps, the quarterly open interest is still much higher than for swaps on on those exchanges. So there there is clear long-term demand for trading

quarterlies. I think the weeklies have been generally um dominated by swaps. And I I I heard some rumors that they even wanted to delist with weeklies cuz they just thought it's bifurcating liquidity and there's no point to them anymore. So Yeah. Okay, yeah, that makes total sense. So whereas the perpet- the the feature of the perpetual swap is basically that you're automatically rolled over um or that it never expires, the feature of the weekly or monthly is that it does expire when you want it to and so in both cases you you you save mental and financial transaction costs. It only depends on what you actually want. Okay.

Yes exactly. Interesting. So, I want to go go back to one thing you said earlier about the early days of BitMEX and how they had to reconstruct their um index multiple times and so on. So, a lot of people say that in um in blockchains and in DeFi in particular, the oracle problem is an intractable problem that can never be solved, but isn't isn't the the problem of index construction exactly the same as the oracle problem?

It's exactly the same. And so, I think to say the oracle problem can't be solved is is clearly is clearly, I think, a very either theoretical statement or uh a very ignorant statement. Because I think the oracle problem is simply how do you bring the needful information from ob- observations to settle financial contracts, right? So, this is already being done in the in in the centralized finance space uh for quite a long time. And the fact that there can be issues with this does not imply that it cannot be

done relatively correctly and usefully for people. I think that that's where that's where, you know, the for a long time there was the idea that well, the oracle problem is intractable, therefore DeFi cannot exist. Yeah. But but if you just zoomed out, you could make the same statement and say CeFi could not exist. But clearly it does exist. So, so so clearly there there's some kind of a strange there's simply it's simply a question of how do you how do you map the how do you map the trustworthiness of index composition in the real world into the uh into crypto or into blockchain. And And

I would point out that you know, for me, I'm aware of many inefficiencies that they have occurred on C 5 index constructions, right? When I was at Deutsche Bank as trader as a prop trader, you know, there are people sitting around all day try try to think about how do we arbitrage index construction, right? There's an entire desk that will say, "Okay, we're going to bet on what things will go into the index. We're going to to abuse the fact that this stock is more likely to enter it and by pushing it up, it is even more likely to enter the index." You know, all these games, right? And you even saw it recently with some of these Chinese companies, where they know how MSCI and FTSE do the definitions. So, they create a shell

company and then they basically do everything that's needed to get it into some of these small cap indices, then to mid cap, and then to large cap, and then I forget the name of the stock, but there was some stock where they managed to get this into one of the biggest indices and no one had done due diligence that this is a real company, right? That that they had actual stuff going on. And the stock went down 99% of the day. And so, the index provider said, "We're going to look into this if this is a real thing or not." And they're going to look into it. It immediately went down 99%. And they looked at And And the And the fascinating thing about that was that that was the week before it was about to get included in a very big index, like in in an even bigger one.

And so, there were some funds that were betting on that inclusion and because the outflow would have been very big because people will be forced to buy the stock if it gets included, then people were betting that this would get included because it it fit all the criteria. But then somehow before that happened, you know, the index came out and said, "Nope, we've changed our mind. We we don't want it." You know, so so, C 5 indices, they're just trying to achieve a goal, same as "We're going to bring in this information and it's going to be useful for settling these contracts, and people generally understand how it's going to go down and what the goals are. But, there's the the I mean, there's always a requirement to zoom out and ask

have we achieved the goal? And if they haven't, then there's there there's always room for humans, I guess, is what I'm saying. Yeah. Yeah. Yeah, I agree. I I think that um that's probably the the going to be the future of of oracles as well, right? So, you have you have a definition of what an index or an oracle are supposed to do, and you have a human looking at data and cleaning the data so it fits the goods. And then they submit the data on chain or in whatever way it is supposed to be used. So, yeah.

And um So, to um to to to end the section on uh I guess that the history of uh perpetual swaps, I I I'd like to move into DeFi. So, in the past in crypto, I I I you could I guess you could say the perpetual swap is the killer product so far in the crypto space. Would you agree with that? Yeah, I think so. Yeah. And I think nothing else has seen this kind of adoption, especially from from retail.

But, it has been surprisingly quiet around perpetual swaps in DeFi. And uh today we want to explore a little bit if that's going to change, how that's going to change, uh what you can look forward to. So, what different approaches exist to create um perpetual swaps in DeFi, and um what what projects maybe uh about to launch in in these categories.

Sure. So, I think DeFi perp swap projects, they generally can be divided into two distinct types. The first being central limit order book driven, and the second being automated market maker or AMM driven. And it's a bit of a mouthful, but uh central limit just means that some participants are posting bids and offers and amending them as the underlying markets move continuously.

So, this would be venues like dYdX, uh upcoming Deri- uh DerivaDEX, a few others in that kind of vein. And the idea there is you're almost mapping the centralized exchange experience into DeFi, right? So, when you trade on BitMEX or Deribit or FTX, that is exactly what's happening, right? There is no automated market maker there, just real people uh putting in bids and offers all day long. And through that process, when the bid of one participant is higher than the offer of another, a trade occurs, right? The matching engine sees that and

submits a trade ticket, submits a fill to each side, and life goes on. So, this approach this approach has some strengths and weaknesses, right? For for uh DeFi. I think the main strength is that it's simple. You're just trying to say, "I'm going to be decentralized BitMEX. I'm going to I'm going to let people exchange risk in this way." And uh that that's quite simple. I think that uh the the downside of that is that you're now taking a workflow from you're taking a workflow from centralized markets uh

and you're trying to put it in DeFi and say that it'll somehow be more useful to people because it's decentralized, right? But you're giving up a lot in the process. You're giving up the fact that well, for one, possibly all these orders are being seen on chain, so they could be front-run, right? You're also giving up a lot of latency. You're also giving up a lot of performance right? Uh you know, a centralized exchange matching engine might be able to handle a million requests in a second. Whereas, there's no blockchain that can handle that.

So, it's it's kind of that that'll be the main challenge. So, for the ones that are going this approach, they're thinking very heavily about how do we scale, right? How do we uh scale, how do we do off-chain matching in a way that people still trust, uh that's still DeFi enough. Mhm. That So, so that approach, I think, is is fundamentally a bet that the scaling will be sufficient to create the experience that is good enough that, you know, users that don't want to use KYC, users that want to be able to just use a web wallet, get in and out, uh they'll be familiar with that. And so, the advantage is is that they'll be familiar with how it works

in uh in centralized exchanges already. So, Serum would be another one that's in that category, right? The Serum would be based on Solana. And they're they're backed by FTX, which would also bring that expertise from central limit order book trading in as well. But again, you have to you you you have to create a user experience where they're saying, "This is actually really good and really useful versus what I can already do on a sex on a centralized exchange." So, the AMM approach, there's a few projects in that space, FutureSwap, uh Perpetual Protocol, Synthetics itself

And and these are saying, "Okay, well, why don't we take the DeFi primitive of the AMM, the automated market maker, and say people when they come into our order book and they and they want to buy or sell, they they don't need to wait for the other side to be there. There will be a pool of capital waiting for them. Right? And that pool will trade against the client, and there will be token holders who are backing this pool, and then they can hedge themselves in any number of ways.

Mhm. So so so this approach is trying to more more explicitly tease out the advantages that DeFi might have over CeFi, right? Which is that it could permissionlessly create this big pool of capital that sits as the market maker. Yeah, I I want to get into the the advantages of uh what uh perpetual swaps could do in DeFi versus CeFi also a little bit later. So Sure. And to to maybe to for our for our listeners so in a previous episode where you gave me an introduction to DeFi in general, you made the distinction between

peer-to-peer contracts or peer-to-peer trading and peer-to-pool trading. Yes. And uh so peer-to-peer to peer is the central limit order book, right? And peer-to-peer peer-to-pool is exactly this AMM model, correct? Yes absolutely. Okay. So I think I guess not a lot of people, including me, have thought about how you actually use the AMM model for perpetual swaps because I'm trying to think um how you would like in in a regular pool where you swap assets, you have this where you basically have a pool of of two different or even more different

assets in uh in a pool and then people can put in one asset and take out another and the ratio in which they can trade these with the pool changes. So, I'm kind of struggling to envision how this would work for for swap contracts. Yeah. I mean, without getting too technical, the the the easiest way to think about it is imagine that you and I and some other people, we opened a BitMEX account and we deposited our money together, right? And we said this strategy will be whenever someone comes to buy, we sell to them.

Mhm. And then we can all see we can all log in and see what the risk of our position is and we also can calculate what percentage of this account we contributed and from there you can imply what your risk is. Right? So, let's say the account, you know, let's say it's just you and me and we shorted 1 million of uh Mhm. of BTC USD swaps. Then you would say, "Okay, well, you're now short 500k and I'm short 500k." Uh Mhm. and you would If you wanted to hedge, you would then go and buy spot yourself.

And you can say, "Okay, now I'm hedged and I have replicated that risk." Uh so, that I mean, that approach clearly works because the the people in the pool are arbitrageurs fundamentally and they're getting their fills automatically rather than having to post liquidity in the order book. So, it's it's it's called a it it you can think of it as a virtual AMM. It's not a real AMM in the sense that it's not like Uniswap where the aggressor into the order book is actually removing some assets and contributing others, right?

There are no real assets being exchanged. Instead, there are positions that generate P&L that are getting given to accounts that are being given to participants. Whether or not they like it or not, these positions get given to you. Mhm. So, Synthetix really is a virtual AMM, right? Because as a as a mentor as a mentor in the system, you have no control over what people will do. You know, you you kind of know what people generally do. But, at any point in time, someone could do something totally weird and your exposure is changing before your very eyes, right? So, Mhm.

that that that means that the you know, the biggest mentors I mean, we're one of the biggest mentors on Synthetix. The biggest mentors, they're calculating what the pool what the pool is doing and what the risk of that pool is and then hedging it, right? And seeing, you know, are there incentives to to mint the opposite risk to neutralize as the the overall risk of the pool? Because some of these protocols will be paying incentives for you to bet against other people that have come in, right? So, let's say there's only buyers, right? Well, then probably all these AMM protocols will have a way to incentivize you to come in and say, "Okay, we're going to actually

pay you to short on our AMM." Yeah. Because if the AMM can attract an equal amount of buyers and sellers, then it actually has no capital at risk, right? It's done a middleman. It's officially become uh the uh peer-to-peer. It's negotiated peer-to-peer right? Uh it's intermediating a peer-to-peer transaction. So, that that's kind of the dream of all these virtual AMMs, which is that it can through through incentives uh balance that risk so that the mentors or the contributors to this uh AMM pool have the least amount of risk.

Yeah. Okay. So, I've I've long, uh, I guess waited for a time to ask some questions about synthetics because I've never really looked into it. But, so the way that I understand it is um, it's not it's not directly peer-to-peer in the like BitMEX sense where like a long and a short meet and then the contract is created, right? So, it all starts out with an exchange and correct me in in a bit if I'm wrong. It starts out with an exchange where you can trade these virtual tokens that represent the P&L, basically.

Um, and and that and via that exchange and via the price discovery the the market can signal that it wants like it it it wants to buy like long tokens or short tokens. And then this other group, the minters, can go, um, deposit collateral and mint against that collateral the token that is demanded by the market and then sell it. And to to tell if that makes sense for them, they they would have to, um, basically see at what price they can can hedge their exposure. Is that roughly accurate?

Uh, it's it's got the main strokes right. It's the mechanism that uh, that that is used uh, it is is a bit different. Basically, what it is is as an SNX holder, you then you then can mint sUSD, right? sUSD is the baseline of the ecosystem. When you mint sUSD, uh, you are now taking on your prop- your your pro rata proportion of the global pool of all assets minted.

So, if everyone else has minted s USD and then immediately bought s- ETH uh, synthetic ether with their sUSD, then if you just mint sUSD, you you are now short ether because the as the price of ether goes up, your debt to the system will go up. Mhm. So, if you if you mint sUSD and then you immediately convert to what everyone else on our on aggregate holds, then you have no risk because you have converted to those assets, right?

So, that that design is is essentially making it so that most people end up minting ether. Uh, or not minting, they they end up buying synthetic ether because they're bullish on ether. Also because everyone else is minting it. So, that also creates some problems for people who, let's say, are only trying to uh you know, mint they they want to mint sUSD but then they want to use it in some other way, right?

So, Mhm. in Synthetix, there's uh, initiatives where they want to encourage people to come in and mint inverse ether or mint inverse BTC because there's also a lot of sBTC that is being, uh, traded into. And so, And before I before you uh, go into that, so uh, for for every um, sETH that exists, there must also exist some inverse sETH? No, no. There doesn't exist this and then because it it's getting a little bit convoluted. I guess I guess the the simplest way I can explain it is if you if you imagine a pool, right? This pool

is not balanced by nature. Who is underwriting the performance of these synthetic assets? Well, it is the people who own SNX. They, in order to for them to claim their weekly fees, they have to get their collateral ratio in line. Right now it's a 600% collateral ratio for instance. So that means that if their debt goes higher than a certain ratio against the value of their SNX, then they can no longer claim rewards. Right? And when they are in the system, they're betting against everyone who's minting synthetic assets, right?

Ah, okay. So yeah, so the the basically the the the other side is not tokenized. Yeah, you could say it's it's virtual or whatever. It's implicitly taken by the the SNX collateral holders, the mintors. Aha okay. Yes. That's right. So so it's a virtual it's a virtual AMM system that's not leveraged. So it's not a perpetual swap system yet. They've announced that they want to go into leverage as well. And so that'll be a another big product in the market.

Something that I I like a lot about DeFi is the ability to like to tokenize any position that you have with any particular depth and then use it in a different one. So do do all of these upcoming systems that you described allow you to tokenize your um your perpetual swap position and withdraw it and use it somewhere else? I think the lower one too for sure because the biggest upside that DeFi perpetuals have is that they can interact with all the stuff that's already in DeFi, right? Whether that's Uniswap pools, balancer pools.

You could imagine stuff where you could imagine stuff where if there's a 50/50 pool of ether and uh a short ETH perp, then that's actually now a crypto dollar, right? That that pool token is now uh equivalent to a dollar. Or you can imagine, you know, on on Wi-Fi, right? You You You could imagine that a that a ETH vault could be a vault where the strategy is to sell ETH to USDC or sell it to uh YUSD, and then yield farm on that YUSD, and then also buy ETH perp to maintain that ETH exposure.

And as the market goes down, they would automatically take their YUSD and use it to re-margin their ETH perps so they don't get liquidated. You know, there's all sorts of things that you can imagine happening. And so I think that's part part of the reason why, you know, we we've been very bullish on DeFi perps as a key primitive because it's really it's really can eat a lot of the things that people currently think of as being indispensable in DeFi. I think, you know, the idea the idea is okay, if people want leverage on their ETH, if they want to be able to borrow against it, well, one way they do it is they put up their ETH and then they borrow dollars. But, another way is they could sell their ETH all to dollars and

then buy a perp, right? And that's actually what CeFi traders do. A lot of CeFi traders, they mostly hold dollars and then or stable coins, and then they buy perps. Uh but, I think one of the big reasons why FTX did so well is that they recognized this. They recognized that, hey, actually in crypto, not everyone wants to hold BTC as collateral. They want to hold their own coin, or they want to hold dollars as collateral, and they just want to trade, right? And so here here you're getting people that same upside where they can hold whatever collateral they want, they can put on whatever positions they want very efficiently.

And so, I think I think in general the ones that are going to be able to benefit more from composability will likely get product market fit much earlier. The ones that are going less composable, more scalable cuz cuz if there's an obvious composability scalability trade-off at the moment in the way that Ethereum has been developing, I think those will take a different route to adoption, but that may also bear fruit in its own way if they're able to create an experience that is really rock solid for users that makes it competitive with centralized exchanges.

Yeah. Yeah, I would like love to go a bit deeper into this idea of creating stable coins or crypto dollars. So, um I have a couple of friends who receive payment in BTC and um instead of selling it, I recommended to them to instead deposit it on BitMEX or Deribit and convert it into a uh high yield crypto dollar basically uh doing the trade that we discussed earlier. And I think I think this has insane potential if it if we get can get this in DeFi. So, I would like to explore a little bit like how exactly it would work. So, in my mind for example, we would

have you would have wrapped BTC and then you would have a short BTC token that that represents uh basically a short position on one of these perpetual swap exchanges. And what you you you could then wrap these two into a new token that is basically a token as crypto dollar, right? Then you could go even further and lend out the original long BTC to earn even more yield. So you would have basically the yield from the short position if the the market is bullish, you get paid positive funding, and you get paid yield from supplying it

on Ave or compound, and you have a stable coin, and it's in one token, and it exists on chain. Yeah. I think that that is in my view a killer use case. So do you think that this is going to crowd out to a degree existing stable coins? Given that this is also trust minimized, right? There's no central issue. Yeah, I think it's not going to crowd out because as as I mentioned before in a different podcast, there is a huge demand for stable coins that can be converted to fiat very easily. Uh Mhm yeah.

USDT or USDC. So it's not a winner take all game in any way, but I do think that these stable coins are very cool. And they could they could make it less appealing to hold shall we say uh algorithmically balanced stable coins. There's less of a need for a algorithmically balanced stable coin with a network effect because there there's just no need, right? You you can easily create a a meta stable coin that is earning the best yield across many things. And I think that gets back to the aggregation theory, right? Which is that, you know, even with all these perp swaps, then you

can imagine once these are all out an aggregator that comes on top and says you know, if you want to short BTC USD or ETH USD, I will automatically rebalance your short across five different perp exchanges. I will find where it's most attractive, and then with with your long leg, I will find the leg that it is that is the best. And then, I will then wrap it for you in a stable coin. So, then, when people are seeing this, they may just deposit a stable coin. This may be a Y vault, right? It may be a vault Yes, I was just going to say that. Yeah, it's a it's basically Y token for perpetual swaps.

There's There's a lot of There's a lot of cool stuff that can be done. And really What it's doing is it's automating what people already do in C 5, but it's making it accessible, right? It's making it very simple to understand. And it's trust minimized. Well, and mostly, people can withdraw it from the exchange, which they could never do, right? They can transact it. You You cannot You can create a crypto dollar on BitMEX, but you can't send it to someone else. Exactly. You can't send your account to someone else, right? You have to take the coin out. You can't give someone your position in BitMEX, but you could do that in DeFi. You could give someone your perpetual swap, right?

Do you think that BitMEX and Deribit and so on will have to enable this? I'm not a believer that DeFi will kill all applications of C 5. I I think that there is a lot of value to what's what what's happening in C 5 from the point of view of of they're already they're already mapping what's happening in traditional finance, but in a very agile and very regulatory arbitrage way. So, in a sense, they have, you know, 1 million orders per second. They create

that experience where you can trade huge sizes against other counterparties. You can trade huge sizes against retail. Uh that that kind of experience I don't think is that disrupted by DeFi actually because that experience has different is ultimately optimizing for different things right? Yeah. So, there will there will there will be parts of the C 5 experience that are highly attackable by DeFi. And I think we've already seen that, right? I think altcoin trading is highly attackable by DeFi because these altcoins are already illiquid as hell on centralized Mhm.

But they're not even that much more illiquid in DeFi. They're probably less illiquid because there's all these people putting it in pools to try to earn some money. So, you you you kind of already have beaten many altcoin exchanges. And I think that's why you're seeing suddenly all the top exchanges list all the DeFi tokens as fast as possible without any fee because they're terrified of the idea that people will don't even bother to deposit. And in fact, people withdraw from centralized exchanges to try to buy DeFi tokens. It's it's a pull factor to DeFi, right? Where Yeah. all the Uniswap tokens going up a lot, people start saying, "Why am I even trading on a centralized exchange where these tokens suck and the ones on DeFi are awesome?" So, I think that part of C 5 is heavily challenged.

I think staking assets as well and governance-based assets as well. You know, if you put it on an exchange, you're you're losing tons of the functionality, right? If you hold certain assets on an exchange, you don't get the rewards. You don't get the right reward rate. You don't get the right governance rate. And all these newer DeFi projects, they're making it explicit that they want you to participate in the governance. They want you to participate in what's happening and they reward the ones that don't and they inflate away the ones who don't care. And through that process, I think it makes them stronger because it it it compels the community to care about what's happening and not just bet on number go up or number go down. So, I think

Yeah. True. those parts those parts of DeFi are are very threatening to what's happening in CeFi. I think that in the perp space the the capital that's currently in cash and carry trades, that's in these kind of things, those can also migrate to DeFi as well. And people will in general get a cheaper cost of leverage than they get right now because they'll their desire to be long will be matched with people who just want to on the stable coins, right? In theory, anyone who's long on stable who just owns stable coins anywhere, they could be a provider of leverage to someone who wants to be long ether or long bitcoin, right? Because that dollar can be used to create

the to to be swapped into a risky asset for them. So, I think it it bridges that vast expanse between the trillions of fiat and dollars in the market and the demand for leverage. So, I think that the last frontier would be the ones where you're trying to compete on scalability uh with centralized exchanges, with these very fast uh market data, you know, these very fast collocated servers. I think that that game is a game that will be the last uh if ever uh that that DeFi would come for.

Mhm. Uh something you you said earlier, um I thought that was super interesting because I never realized that DeFi tokens tend to have an ongoing proof of work for the user. Well, it's not just the moment they are mined initially or created initially that there's a proof of work, but if you are a holder of these tokens and you don't want to get diluted, then you have to perform a a constant proof of work basically for the entire duration of the the asset. It's just kind of crazy.

Um Yeah. So um about another trade-off between um DeFi perps and uh CeFi perps. So, we saw uh Deribit announce KYC. So, BitMEX announce uh KYC. And um well, isn't that one of the major benefits of DeFi and one of the major pull factors in the future? I think for a lot of users that want the centralized exchange experience, but without KYC, there's going to be a lot of value in the DeFi perps that take the first approach that I

mentioned, which is the central limit. That means that they can market make, they can work a bid, they can work an offer, they can get that same experience that they get uh Yeah. elsewhere, right? So, so I think there is a big set of that. I think that the exchanges where uh So, so I think one thing that uh is interesting is people are quite okay KYC'ing if they think it's worth it and it's trustworthy. So, I think when when FTT did their to- when FTX did their token and you had to KYC to be able to get the token, all of a sudden people were like, "Oh, obviously I'll KYC for this. It's It's easy money, right?" And and I think I And so, I think that there is

Yes, there is the demand to not KYC, but at the same time there's also people don't value their KYC very much in dollars. So, it's it's kind of a It's kind of ultimately still about comfort, still about trust, and still about uh cost. So, that's why I think centralized exchanges can They compete from the fact that for one, they they uh they have specific client bases. Some of them are more experienced with what's happening in the traditional CFD space, right? For instance, an exchange like Bybit their their DNA is much more from the traditional retail FX space. So, they understand the affiliate marketing, they understand

that kind of kind of a flywheel. And so, you know, there there's a plurality of business models, I think, that that can exist uh in this ecosystem. And I think that there's I don't see many reasons why it's winner take all in in any case. Uh but, I do think that there will be some parts of perpetual trading that will be heavily challenged. And and I think one interesting example of this is just copy trading, right? Or asset management.

Uh I think the a really cool thing you can do now with perp swaps is, you know, let's say some guy tweets that he's a really good trader and all this kind of stuff. Well, now, like in theory, he could create a portfolio, even cooler if it's a shielded portfolio, where he is putting his positions out online and people can shadow him. And let's say if he's good, you know, the track record is all on chain. And and people could then invest in something where they get the signal at the same time that he does. So, so they get the same fill, right? Uh all these kinds of things, they can't really happen very easily in a centralized exchange space, you know, you you have to give the guy the API keys, you you'd have to hook up his algo

to your thing, and you have to trust that it's going to work. I think that uh asset management uh on chain is going to be pretty big. I think that uh you know, Y vaults are kind of an example of that, but there will be a lot of other types. Yeah. I think that uh I'm more excited about the the sort of non-zero sum convex opportunities that will be created by the explosion of design space from DeFi as opposed to saying, you know, what what things in CeFi will will will be dominated strictly dominated by these. I think net it improves the whole experience of both DeFi and CeFi.

Mhm. Another example of that would be, you know, in CeFi right now your collateral earns nothing, but when you integrate, you know, I could imagine a a CeFi exchange integrating like the DeFi perp, right? It could include it in the aggregation. There could be a market maker that brings that in. And now mean, you you don't even need to do that, right? You can just supply the collateral into Compound and Aave and supply the collateral. Yeah, you you could also do that and then and then you could take one step further and and kind of do uh kind of do more sophisticated strategies or or do more sophisticated bridging of liquidity.

So, Mhm. there's there's always going to be like a meeting in the middle that if if you will because centralized exchanges what they have is a as a strength is a captive flow, right? They they have people that want to trade there because they trust the brand, they understand the user experience, the client experience, they know that it's safe. They want to be KYC'd, right? They don't want people to be able to randomly withdraw through a web wallet. They they want there to be a very hard withdrawal uh process. Uh Mhm.

as if it's an institution. Uh Yeah. You know, on some exchanges, you know, we we have protocols where I have to get called, people have to get called for that withdrawal to go through. Uh Yeah same. It has to be a video It has to be a video call. So, so So, I So, I think that many participants, when they're handling large amounts of money, they prefer this kind of thing. And other participants, uh if they just want to click around, they they honestly might go fully to a DeFi solution. And there's a whole range of mix we Um Okay, so um by the way, I also um explored some of the ways that centralized exchanges can hook into DFI protocols such as compound in in an article called uh the great race to

crypto banking. Uh I'll be putting that link in the show notes. So, I have a final um topic that I'd like to explore. Um We both have in the past argued that the BitMEX insurance fund was too large and um I have two questions about that. So, first um it earlier in March this year there was this huge um day where Bitcoin crashed like two two two three and a half K on BitMEX and some analysts have suggested that the insurance fund would have been drained in the single day had BitMEX not internalized the the toxic flow. You remember the event, right?

Yes. I did. And um did that change your mind about the uh like the optimal size of the insurance fund? It definitely definitely made me think that it's not the worst thing in the world for one of the biggest exchanges to have a bit more of a larger one, right? I think that the the way that BitMEX works with its cascading liquidations um it's uh how do you say it? It definitely it definitely benefits large participants to know that there is a big insurance fund backing it.

But but I think that the market will come to more capital efficient ways to solve this problem, right? I don't think that having a gigantic pool that just sits there and and uh I don't know that this is the best approach. I I actually think the tokenized approach uh ultimately is better where you can tokenize the future revenues of the system. Something like that. But Yeah. In any case In any case, I think that you know if BitMEX had partial liquidation, if they had some kind of a non-cascading function, I think I think it wouldn't even have gone down that much cuz it was really all BitMEX led after a certain point,

right? I think below 4.6, it was entirely BitMEX led. BitMEX went over $1,000 below everywhere else and it was all arbitrage, right? People were just trading BitMEX on the others. 100% BitMEX, yeah. Yeah, I mean we were very fortunate that day because of the way that we were positioned, but you know there were people trying to to trade market neutral that got blown out because they were long BitMEX, short elsewhere and then they get wiped out on that leg. So, you know it it's I mean it it definitely gives shorts that uh On the day, you know, they they feel good knowing that the insurance fund is there. But on the other hand, if you look at every other exchange that day, they didn't have any problems with their insurance fund uh or any huge problems.

Uh and I think part of the reason is that they liquidated in a slightly different way and the prices were more orderly, the move down, right? Yeah. So, I think I think that BitMEX definitely took a big hit after that day in terms of client trust uh in terms of having their market go so low and stay so low for so long. I think that um I think that big insurance funds, I don't I don't anticipate there being another one. I think that BitMEX being the first, they managed to accumulate this when when clients sort of didn't really care uh or didn't really calculate how much they were losing to to the slippage.

And now that there's a lot of other ways, you know, people now know they can get partially liquidated there of it on on FTX where they don't lose their entire position just because it starts going into a margin call. I think people are realizing that like it that like this is real money that they're putting out there that that they're losing. So, I think that uh I don't expect big growth in any other insurance funds. Yeah, and to share our own book here once more, so we also had an article on the design space for insurance funds and liquidation mechanisms and I I'll also be putting that in the in the show note.

So, my my second question is and that's something we did not explore in the article but that is nonetheless very interesting to me is how does DeFi change the design space for insurance funds and liquidation mechanisms? Is there anything that we can do now in DeFi that maybe we couldn't do that hasn't been tried in the CFi space? Yeah, so there's a few things you can do, right? That that created embedded backstop to the system while be more capital efficient. I think one is you could you could mint new exchange tokens whenever insurance fund goes negative.

That's kind of a cool concept, right? Because you're using total cash flows uh and meanwhile everyone else who's trading pays a cheaper price, right? And one protocol that does this is Maker. Right? Yeah, Maker exactly. They do this. I think another another good way is through uh staking, right? People who stake the exchange token, they bear the risk of the insurance fund going bad, but they earn more of the fees. So, so in a way they're kind of like the the equity holders.

I think I think there's a lot of interesting ways you can handle that problem. Ultimately, what you're trying to do is trying to do a risk insurance, right? You're trying to insure against a move that wipes out um participants to the point that you cannot pay some other participants. And that gets us back to the AMM discussion, which is that how do you ultimately ensure that uh the flow is relatively balanced and that you're the people who are the people who are uh needed to be able to back up the payments can actually make those payments. So, I can imagine there being some synergy with insurance protocols, right? Where there's people that put money in passive pools that say, "I also want to do uh credit delegation or I also want to

do some kind of a delegation where I want to earn a higher yield and I'm okay being slashed if this exchange goes down or this exchange's pool goes down, right?" Yeah. Think like a really cool development we just saw was I think I think Aave proposed it a year and improvement proposal, which was to allow credit delegation in Aave. Yeah. And you could imagine you could imagine like in the future like other types of credit delegation regarding insurance funds for for perpetual protocols being very much about how do we how how do we access that vast pool of capital that's sitting there and wants to earn more money, right? Because that because that capital is is relatively

risk-seeking, right? It it it's apparently shown the interest in farming random coins, right? 500 million of this stuff goes into random projects that that are unaudited every week. So, clearly there's risk tolerance here. And hopefully some of that risk can be harnessed in a sort of market-efficiency way that that uh is a win-win. Right. So, it doesn't just go into the next decentralized Ponzi scheme, but instead it goes into the the also risky and high-yielding uh insurance fund of a perpetual swap protocol, but that also yields something for the greater good for like DeFi the DeFi space as a whole. Yeah, makes a lot

of sense. Yeah. Yeah, so um thank you for this amazing discussion. Uh thanks for thanks for the talk. It was a lot of fun.