Doug Colkitt is an ex-high frequency trader, ex-MEV bot searcher, and founder of the decentralized exchange CrocSwap.
In this episode, we talk about all three. We begin with high frequency trading, where Doug walks us through the differences between maker and taker strategies, why queue position is so critical for makers, and why volatility is a high frequency trader’s best friend.
We then discuss Ethereum-based MEV strategies. Doug explains what MEV is, how the architecture of the Ethereum block chain allows it to exist, and a high level topology of the different types of MEV strategies that exist. He also explains how the game theory behind MEV changed dramatically with the launch of Flashbots.
Finally, we talk about his new decentralized exchange CrocSwap and its primary innovations, including dynamic fee levels, identification of toxic flow, and vaults that enable KYC.
I hope you enjoy my conversation with Doug Colkitt.
Transcript
Corey Hoffstein 00:00
All right 321 Let’s jam. Hello and welcome everyone. I’m Corey Hoffstein. And this is flirting with models the podcast that pulls back the curtain to discover the human factor behind the quantitative strategy.
Narrator 00:20
Corey Hoffstein Is the co founder and chief investment officer of new found research due to industry regulations he will not discuss any of new found researches funds on this podcast all opinions expressed by podcast participants are solely their own opinion and do not reflect the opinion of new found research. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of newfound research may maintain positions and securities discussed in this podcast for more information is it think newfound.com.
Corey Hoffstein 00:51
Doug Colquitt is an ex high frequency trader X MeV bot searcher and founder of the decentralized exchange Croc swap. In this episode, we talk about all three. We begin with high frequency trading, where Doug walks us through the differences between maker and taker strategies. Why Q position is so critical for makers and why volatility is a high frequency traders best friend. We then discuss Ethereum based MeV strategies. Doug explains what MeV is, how the architecture of the Ethereum blockchain allows it to exist, and the high level topology of the different types of MeV strategies that exist. He also explains how the game theory behind MeV changed dramatically with the launch of flash bots. Finally, we talk about his new decentralized exchange Croc swap and its primary innovations, including dynamic fee levels, identification of toxic flow, and the vaults that enable KYC. I hope you enjoy my conversation with Doug Culkin. Doug, welcome to the show. excited to have you here. There’s a bit of a growing theme this season, which is talking about high frequency trading. So I’m super excited to have you on as an expert guests to walk me through it. So welcome.
Doug Colkitt 02:10
Super excited to be here. Thanks for having me.
Corey Hoffstein 02:13
Why don’t we start with your background. I’m really excited to get into things that you’re working on today. And you just clued me in, you’ve got a big launch coming up that I want to get to at the end of the episode. But let’s start at the beginning. walk our way towards how you got to where you are today. I
Doug Colkitt 02:29
started my career in the late 2000s After school went to work for a company called Citadel at the time who was more known in the hedge fund space. When high frequency was still emerging. Started there initially was in a rotational program got into the high frequency desk, I thought it was really interesting. The stuff they were doing work there for a few years as a quant researcher, so combination of working on signals, they’re working on various strategies, I spent most of my time there in their Asia Pacific equities strategy. So there’s a lot of interesting markets. But to learn a lot, let’s put it all in the early 2010s. Then I kind of decided to go out and see if I could build a whole high frequency system from scratch. So a bit of an audacious goal at the time, still kind of kid in my 20s but decided to build a high frequency trading strategy on the CME focused on taking liquidity instead of market making driven by high frequency alphas did that for a few years, that got more competitive by the mid 2010s. So hit some other constructs for a bit but transitioned over to kind of randomly market making equities in perky for a bit This is right after their exchange when electronic. So figured it was kind of a growing market at the time and a gold rush there when people were moving in and kind of their systems were getting modernized. So did that for a few years, and then kind of randomly in 2020 read some cool stuff about this concept of MeV. I’ve never really done much in crypto before but I thought it was super interesting. It’s kind of a side project but quickly kind of grew to take over. So the MeV searcher for a bit. And then side it. Rather than trading. I kind of wanted to build kind of these new decentralized exchanges and protocols and not just trade on them. And the space was still growing a lot of innovation. So then I started my project Croc swap in 2021. And I’ve been working on it since and yeah, we should be launching this upcoming week.
Corey Hoffstein 04:33
Oh, congratulations. I should point out that by the time this episode gets released, it will already have been released. We’re recording this in early April and this season won’t come out until May. So in our time machine, congratulations on the launch. Thank you. Lots of fun stuff to unpack there. Let’s start at the beginning. Post citadel. When you initially struck out on your own, you decided to trade equity futures rather than equities, which is what you had been doing at Citadel, curious as to what the rationale behind the change was, and whether you can comment on similarities or differences in trading the two different seemingly very highly related asset classes.
Doug Colkitt 05:19
The biggest reason that focus on futures instead of equities was well really, I guess few things, but they’re pretty closely related. So with at least equities in the US, there are multiple venues of NASDAQ Niazi, that’s direct edge, kind of all these different venues. And that’s fine, even getting into the dark venues, the off exchange stuff, neutralizers everything there. So if you want to really have a decent operation in equities, the infrastructure footprint you have that from day one is pretty expensive, right? You have to be co located all these venues, you have to have market data agreements with all these venues, direct market access everywhere, as well as being able to send data price discovery is happening at different venues, you have to be able to send data quickly, which also gets expensive sending from one data center to another kind of managing all that plus or regulatory issues with like the markets have to stay aligned you have to be aware of, and BVL futures is a lot simpler, because almost anywhere the futures trade, they just trade at one venue, right? I can’t go take a long position at the CME and then even if there’s a similar futures somewhere else, I can’t move it from the CME to the other venue like I could buy Microsoft on it and sell it on NASDAQ. So because futures only trade at one venue is just a lot easier, you only have to be co located in one space. So as a kind of a woman show, the infrastructure was a lot feasible to get into position there. The second thing is mentioned kind of that dark flow, internalized flows. So the challenge with equities as a lot of the nontoxic flow doesn’t even hit the lit markets to begin with. So a lot of the game in equities is actually getting access to order flow upstream before it even gets the market. So a bit of a barrier there compared to someone who just wants to create an exchange.
Corey Hoffstein 07:02
In our pre call, you said to me, quote, to trade high frequency you need signals that change on a high frequency basis, is one of those quotes that I would categorize as simultaneously being incredibly obvious, and yet somehow both potentially profound and insightful at the same time. What do you think this quote ultimately means for the types of signals that high frequency traders can use?
Doug Colkitt 07:29
The issue is, if you really want a strategy that turns over quite a bit that holds positions for short periods of time, reality is right in the signals you use, or whatever, using some structure to change as to actually think frequently enough to match the cadence that you want to trade. Even if you’re trading on an hourly basis, there’s a lot of information that affects the price of any asset that kind of moves, really. So now, the good news, social type stuff, not to mention, like actual economic data, all that stuff. So even once you go beyond the high frequency round, even to the realm, things open up a bit. But in that high frequency realm, if you’re targeting, say, on the order of seconds, or maybe even up to a minute, there’s just the number of things that actually change at that cadence is very limited. And really, what it comes down to is the only things that are happening on a stock from a future whatever on a second basis is the trading that’s going on in the market itself. So in that sense, when you’re trading high frequency, it’s really interesting, you don’t really care about the asset itself. And that’s why firms that are good in crypto, high frequency are the equities good and futures very easy to move over. Because all you care about is kind of this market microstructure that it tends to be pretty similar from place to
Corey Hoffstein 08:42
place. So volatility ended up drying up in the equity futures market post GFC, that 2012 2013 period, love for you to first comment on why that changed the nature of your ability to be profitable. But I know that was also the transition point for you to end up making markets in Turkish equities a couple years later sort of gave you the catalyst to change what you were doing. So when I talk to high frequency traders, there’s this big line in the sand between making versus taking. So as we maybe transition a little bit from what you were doing with equity futures to what you were doing in Turkish equities. Can you define what making versus taking is and how they differ from an alpha perspective? So hitting you with two questions, one, maybe just talk about that regime change and then to high level, what is maker versus taker
Doug Colkitt 09:36
in terms of regime change? I think one reason it was easier to be kind of a smaller shop or even even a one man show in the early 2010s Right after the GFC was because when volatility is really high, the dimensionality of the strategies you’re running goes up a lot, even just down to the level of the number of events that happen every one second on average. was a lot higher in 2010, than it was in 2015. Everyone’s kind of building similar models, right? Blood pressure, you might have order flow, you might whole movement of correlated instruments between different books, you’re not using exactly the same model, everyone kind of has their own variation, their own spin, even just like the way you implement might be subtle details that change it. So when things are very active, if you have a model that’s slightly differentiated, but maybe not as good as one of the top players in the space, there’s more room for that model to be able to extract marginal alpha at the price discovery, when markets get a lot more quiet. And particularly when volume goes down relative to how much high frequency activity there is, things come a lot flatter, in terms of how many different ways can you do the same thing. And in that respect, it’s kind of like the apex creditors feed first, and whatever’s leftover is a lot more likely to be toxic for you. So one of misconceptions people think in taking, you don’t have to worry about toxicity at all, that’s only something that makers have to worry about adverse selection. But even in taking, especially when you’re competing for the same events and latency, competitive things, there is adverse selection, in the fact that if 10, other firms are trying to hit this alpha all at the same time, you’re a lot less likely to get filled. If your model is wrong, you’re the only one. So even in taking adverse selection is a problem is a lot more of a problem. When volatility and volumes dry out, which kind of drives secondary or smaller players out of the market and think is big news. And you saw some consolidation in the 2010s.
Corey Hoffstein 11:33
So you ended up moving to market making in Turkish equities. Again, I was hoping you could just draw a really clear definitive definition around taking versus making what does that mean for someone who’s never operated in the high frequency space? How do they differ from each other?
Doug Colkitt 11:49
Well, I mean, just take a step back. There’s an order book, there’s resting limit orders in the book, if I want to trade at any given time, there’s a bid, there’s an ask, when to buy PDF, when to sell, I received the bid the bid slightly lower than the ask right? So the idea with taking is, I’m going to trade right away. So I have some signal that I think is number one very quick time decay, I don’t have time to wait to try to get filled, I need to fill this right away, because I think it’s only going to be here for a very fleeting period. And number two, it has to be big enough where crossing the spread painless transaction cost justifies it in terms of signals. So taking looks a lot more like stat are from that perspective, right? I’m trying to fit signals. I’m trying to make sure my signals are accurate. It’s not just enough that they’re directionally right, I have to get my magnitude, right, because I don’t want to fire off trades and pay transaction costs when the expected value isn’t high enough. So in the sense that actually monetizing and taking strategies a lot simpler, right? I’m just trying to Okay, check my signal, see if it’s big enough. And yeah, I manage my inventory, trading that but that’s not super difficult to do. And the nice thing about high frequency is your Sharpe ratios are high enough the risk, and like hedging isn’t like a super big insurance, your main thing is, I need to be positive on every trade I make making right is you’re on the other side of that you’re someone who’s resting in the limit order book, you’re providing liquidity, you’re putting in bids and asks out there. And by default, you win. If somebody comes in and sounds to you at the bid, and you buy at the bid and you sold the asset, you make money kind of by default, which seems simple. The problem is the default doesn’t usually happen, right? Because people who are taking are kind of aggressive and not stupid. So with making the monetizing, how do you monetize? It gets a lot more complex. And then the other thing is, with order books themselves, there’s kind of all these complexities. Where do I place my orders? If the bid is at $44.02? Should I improve it to $44.03? can I improve it, maybe the bid ask spread is tight, and I can’t because the tick sizes are there already wondering why. Or maybe I joined far away and wait for the price to move to me. And then there’s this whole issue around composition. So each level has a first in first out queue on most order books. If I’m at the front of that queue, I’m probably gonna get filled with very high certainty. So therefore there’s less toxicity, less adverse selection. But I’m at the back of the queue, the fact that I’m getting filled, probably means that someone has sent a very big market moving order, and that’s very bad. So you still use outflows on making two and they still add a lot. But it’s also right, because alphas are kind of time decaying. They don’t add as much so there’s all this complexity around how do you actually monetize a strategy, but maybe the alphas are less competitive.
Corey Hoffstein 14:30
I want to dive a little bit deeper into that cue position concept because as someone who trades much slower frequency alphas in my day to day job, that is not even something I’ve ever thought about before I get there though. I’d really love to sort of like distill this maker verse taker, and you tell me if this is an overcomplicated way of explaining it, but my interpretation is almost taker is you are using alpha signals with the expectation that the market will continue to move in the direction of which you’ve crossed the spread. To justify your price of the spread versus maker, the hope is that when you get filled, the market will either stay flat or revert so that you could offload your position and not have to cross the spread to unload. Is that Is that a fair way of sort of separating them?
Doug Colkitt 15:19
I’d say that’s definitely fair. The only caveat I would add to that is, you say like the market will continue to move that kind of implies that there’s kind of this continuation of a trend, it’s actually, at least in my experience in high frequency, whether you win or you lose is like really, within the first 100 milliseconds of what happens. So if I’m taking, I want that level to crumble right away. And if that level doesn’t move right away, right after I get it, behind me, usually it’s a bad trade. And similar to making, if I get filled, and that immediate field doesn’t push through the entire level. Usually, that’s a good trade. So it’s almost instantaneous, instead of this continuous, I mean, obviously, the market trades continuously, but like these events, kind of get decided very quickly.
Corey Hoffstein 16:07
Let’s talk about those layers, because again, unless someone’s involved in the market microstructure, an actual execution, they might not even be aware that these layers exist. So in talking to you, it sounds like cue position is almost sort of everything a hugely important aspect of the alpha you generate on the market making side. Can you talk about what cube position is? What are layers in the order book? How do they operate? Why is cube position? So critical? And why does it lead to this behavior? Where market makers have resting orders in the limit book? And then as they get closer and closer to the top, they cancel and pull them? What is that behavior all about? So I was hoping maybe you can just sort of shine some light on this market microstructure aspect.
Doug Colkitt 16:53
Before even that, one thing that you should distinguish Is there really two types of books on different instruments one is what would be called like a thick book. And that’s basically where tick size is very large relative to how much people want to trade. So a lot of instruments are thick books like the s&p E Mini is thick book. A lot of US equities are thick book definitely in Turkey, where prices were like five, literally. So time is trading and six for one penny things are thick book. So when the tick size is very big, that’s almost like a price floor. Right? If you think of the bid ask as like, what’s the price of liquidity, when size is set very large relative to where people would naturally want to put the bid. So it’s almost like a price floor ratio of liquidity, it becomes you can’t compete in terms of setting a better price, because usually the market is the bids at like $44.01, the answer is at $44.02. And the minimum tick size increments is a penny. So there’s kind of no way to improve on that. So market makers stopped competing in terms of price cuts, because they can’t in those revisions. So with that being said, what happens is pretty much every possible price has some order set at it, because there’s just not enough granularity in the market. And there are different levels. And what becomes really important is if I can make that bid ask spread with like high certainty, it’s very, very profitable, because almost like the bid ask spread is set artificially, too high. But what determines whether you get filled or not, it’s pretty much your position in the queue itself. There’s a first in first out system, so I’m the first person to put in an order at $44 on one cent, that means the very first marketable trade that hits that price, I get filled. So there’s this adverse selection aspect of as I go further back in the queue, the likelihood that I get filled is much, much lower at any given time, which means that the fact that I get filled and the queue is really long, that probably means someone has like some market moving information or signal or they’re just working on a big portfolio. And that probably means immediately after I get filled, the price is going to move against me. And I’ve kind of lost money. First, I’m on the front of the queue, even like these tiny one lot orders I’m getting filled against. So there’s that aspect of the adverse selection. And then what you mentioned is why do people put orders further down, like further away? And why did you cancel so much, it’s almost like you would solve, you can kind of think of it as an option. So I put a position in the queue. If I don’t get filled right away, I’ll be okay. And maybe I’ll move up the queue, right. So the things in front of me get filled. So it’s kind of like I’m sitting there, I’m taking risks, and I make it filled right away and probably a bad feel. But the longer I sit there, the more I move up the queue and the more profitable any field will be in expectations. So when prices are far away, and you have this stick book scenario, the probability that prices move like two ticks, in a very short period is pretty small, just because ticks are large relative to volatility. So it’s almost like a free option. So if the bid is $44.01, I’m sitting at $44. I’d like a free option. So I’m holding this Q position and hopefully it moves up I get there. But if it does move up, and I am at risk of getting filled, and I don’t like my cube position, which is probably 90% of the time, 95% of the time, it’s free option, I just held it, I’ll cancel right away. I don’t pause it anymore. Because there’s really no cost to putting an order or canceling. Maybe there’s some order to trade limit, like minimums, but they’re usually pretty loose. People use this as a free option, basically, the whole key position.
Corey Hoffstein 20:23
My naive interpretation here is that it comes down to maximizing your fills without the market rolling over you and your spread. So what I need you to do is clarify something for me, you hear this idea that market makers make the most money when volatility is high. With these super high volatility regimes, market makers are just printing p&l. That seems like exactly the type of environment or period where you could easily get run over in a single direction. So I was hoping you could reconcile that. For me.
Doug Colkitt 20:56
It’s kind of a paradox. And it’s definitely non intuitive. But yeah, when volatility gets super high, and markets start moving really fast, it’s actually market makers can generate a lot of money. And the caveat being is they can probably only generate a lot of money if their infrastructure and their systems are good enough to keep up with the market. So like class samples, like a flash crash, like things happen, market moves way too fast, a lot of people just haven’t built their infrastructure to process data that fast, especially the non high frequency traders, maybe people working on like a V Whap order or something, their view of the market is totally out to date, they’re just kind of firing off random orders, they don’t know what the price is, they think they’re putting a limit order and the price has moved a lot, it’s usually not the market moves straight in one direction bounces around like crazy, just because all this random order flow is common. It’s actually when markets get crazy like that, and you actually look at the order flow, the order flow is actually not very toxic, at least not as much as you’d think it’s much less toxic than you think. Because of that, it’s actually not too bad. And you’re and maybe the market moves through you, but it bounces around a lot and moves back. And in addition, right a lot of market makers can’t keep up so they turn off so you have less competition for Q position. And then the third aspect is like a lot of these books are thick book, they’re always one foot wide. But oftentimes, when you hit like these flash crash type events are very fast market spreads will widen out, because again, that goes back to like market makers are just gone. People who are trying to maybe passively work and execution order are not in the order book. But some people provide limit orders. If they’re working on V whap, algo to get better execution, they’re gone, they’re wiped out, they can’t keep up with the price moving. So actually, because of that, you might get a situation where a tick that’s normally a penny wide, almost all the time is like 510 cents or whatever, whatever wide. So even if you get hit more, you’re making a lot more in that script. So I’d say those are the main reasons.
Corey Hoffstein 22:49
All right, we’re gonna make a big jump now to the world of crypto. You mentioned after market making in Turkish equities, you started doing something called MeV, which for I’m sure a bunch of listeners, they have no idea what MeV is. And this is something you spent a couple of years on. So to set the table for this conversation. Can you first talk a little bit about the architecture of the Ethereum blockchain? And in that context, what is math? What does it stand for? How does it
Doug Colkitt 23:20
operate? With a theory or really any blockchain? What happens is, you have block producers. So some party that at any given time can produce the next block a block contains a sequence of transactions, the block producer pretty much has total freedom to sort those transactions, how they want. So Ethereum, for example, block times 12 seconds, so every four seconds, you get a new block of transaction, whoever’s the validator in that period, or if it’s a proof of work system live, or the Miner is that mines, the block can decide how to put those transactions inside that block. So that’s how the basic blockchain works. That being said, there’s interest rate, how do those transactions get to the validators or the block builders, and this is a very nowadays, but traditionally, there’s what’s called a mempool. So it’s a peer to peer network. It’s basically like Bitcoin are like a fileshare network, peer to peer network of what’s called a gossip network, I sign a transaction I broadcast out to the network, everyone’s broadcasting these transactions back and forth. And then a validator can go out, look at the transactions that pay the highest gas gas is basically the transaction, the amount you’re paying the validator, a miner, put your transaction in the next block. So traditionally looks like validated goes out into this peer to peer network or this mempool gets the transactions that pays high enough, puts them in the next block and then spits out the next block. That’s how it works. And the final wrinkle is on Aetherium or other smart contract chains. You have gaps, this would be centralized exchanges, which probably get into that but like the ideas, right, you can trade swap assets. So it’s pretty interesting in terms of how does this differ from a traditional market the Biggest thing is, a lot of transactions are visible to everybody else, all the players in the market before they actually go to the market. So that’s different than if you’re always at the CME or an exchange, or NASDAQ, I send my order, before anyone else can see my order, it’s atomically put into the book, it interacts with whatever marketable liquidity is there. And so no one else can really see my order. And that’s why they’re already resting in the book or was executed. In this right, I can see the order I can see, okay, or you put out a swap, you want to buy, Kenneth Areum, for $2,000, or whatever. So I know your orders there, I can have a good idea how it’s going to interact with the decks or decentralized exchange, and I can kind of predict what’s going to happen. And therefore I can hopefully be someone that can make money. With me knowing ahead of time, what’s going to happen, there’s different variations of how you can actually use that.
Corey Hoffstein 25:48
So the core idea of MeV, then is basically looking at the existing mempool of a block that hasn’t been struck, and then taking actions to try to profit on the standing transactions in the mempool.
Doug Colkitt 26:03
Yeah, exactly, I’d say right, there are deep analogies to high frequency trading in the sunset, high frequency traders in traditional markets are the people who are probably most skilled at this, like very low granularity sequencing game or like how things go through and understanding what’s going on a very low granularity, and MeV searchers are similar, not really moving tons of money, or at least like in one direction, but trying to get in and making pennies off of how things are sequenced, or at a little granularity level.
Corey Hoffstein 26:31
Just for the sake of completeness here. What does MeV stand for?
Doug Colkitt 26:34
Oh, MeV actually stands for minor extractable value. This is ultimately the block builder. And this is before the theory and was proof of stake and those reminders to build the block for therefore, ultimately, it’s the miners who can extract this value. Now back in the day, it’s actually the miners were pretty naive. So most of the value went to actually third party searchers didn’t really have any relation. And then the miners kind of got wise and got smarter about extracting more value on there. So
Corey Hoffstein 27:03
can you classify maybe the primary MeV strategies and provide an example of how each of them works,
Doug Colkitt 27:10
there’s longer tail, there’s more niche strategies, but of the big three, I’d say it’s front running, back, running and steady. So go through each one. So front running, is probably something has the least analogy to high frequency trading, because in most cases, like I said, orders are atomic. But the idea is I can see an order that’s coming into a deck. So typically, orders come in with a certain slippage, so or you might sign a swap saying I want to buy Aetherium, the price in the decks is 2000 right now, but I will pay up to 2050. And I can see that I say, hey, the price index is 2000, I know you’re gonna pay 2050, I’m gonna go out, buy up on the theory, I’m ahead of you front run new, so I’m gonna go to the decks first I’m gonna buy up some theory and then put your transaction after mine. So you pay too high of a price, you’re paying 2050 instead of 2000. And then I’ll go back and I’ll sell you. So I go out, buy some and then I basically sell to you. But I mean, really, you can move the pool, there’s different mechanics around it, but that’s the idea. I see your trade, I want to go out and get the worst price, and then I profit from that. That’s trunk running. The second major category is, quote, unquote, back running. And that’s basically where I see okay, you’re gonna go in, you’re gonna buy one deck. So let’s say you’re just going to swap. And let’s say look there sushi swap over here in the ether, have an Aetherium USDC pool, I know you’re going to buy it uniswap You’re not trading it. So she swapped, you’re going to move the price. And now there’s going to be an arbitrage between the two. So I go in and I say I want to make sure right after your swap lands, the price is going to be dislocated. I want to be the very first transaction after your transaction because you push the price at uniswap up twice, so I’m gonna go buy it sushi swappers cheap, sell the uniswap words expensive and kind of walked in this quote unquote, atomic arbitrage profit, it looks pretty interesting there is compared to Trad fi where even if I’m doing kind of a similar thing between like Nike and NASDAQ, that trade isn’t guaranteed, right? Because I might get filled at NASDAQ, but not fulfilled that nicely. With this, because it’s a smart contract chain, I can put in a transaction, I can run all these things. And then I can check at the end, let’s make sure this actually went through. And if I didn’t, I’ll just revert my transaction. And we’ll go through. So it’s really interesting. There’s actually like a purely risk free way to make money and you can even stack on top of that you can do a flash loan, where I borrow a bunch of money at the beginning of the transaction, I don’t need any collateral will abort the transaction if I don’t pay it back, and then I pay it back again. So technically, this is like really the only way really almost anywhere where you can make money trading without any risk or without any capital besides maybe like the gas costs to actually pay the transaction but that’s kind of like just a really fascinating aspect of it. And then the third strategy is like stat are not might look more like traditional setup in a market where maybe the catalyst isn’t actually what’s going on in the blockchain. The catalyst is something going on somewhere else, particularly these centralized exchanges move a lot faster than the blockchain Aetherium only moves every 12 seconds. So if the price of Aetherium goes up by Nance, I know probably the price of Aetherium on the uniswap pool is too cheap or at least is too cheap. If on the first transaction in the next block, I want to hit that sale price, and I get filled, and there are pure arbitrage isn’t necessarily possible, because I would have to move it from the blockchain to binance, which obviously you can’t do on one transaction, but maybe I know I can make money in expectation. So I know if they consistently buy ether when it’s cheap at uniswap, and sell it when it’s expensive. I’ll make money and expectation over time, probably with a very high Sharpe ratio. So those are, I’d say, the three core strategies, and you can obviously do variations on them. But those would be the main things that searchers will look
Corey Hoffstein 30:47
at the risk of sort of beating a dead horse here, miners during the proof of work arrow could do this, because they could inject any sort of transaction they want into the block that they are compiling. Why is it possible for MeV searchers to do this? Why can you suddenly inject your transaction before or after mine? Create some sort of sandwich attack? Is this not just a FICO model that the first transactions in or the first executed? How does the Ethereum blockchain architecture actually enable these types of strategies?
Doug Colkitt 31:21
So the interesting thing is how the block is constructed and where transactions go, that’s all part of the protocol itself. So any miner or validator is free to do whatever they want and put whatever transactions they want. So before MeV, was a really big deal, what miners would do is something pretty simple. They’d say, let me look at all the transactions out in the mempool, who’s paying the highest gas price. And then if it’s tied, whose transaction did I see? First. And miners weren’t really aware of MAVs, they were just kind of running this naive algorithm is kind of part of the standard theory and client software. But there’s no map, this is a greedy algorithm. And it pretty much works correctly and maximizes your profits as a miner. So before my neighbors were really aware of if I could exploit the fact that miners just use this kind of naive algorithm. And I can say, Okay, I know a miner is always going to take the transaction with the highest gas fee, and therefore I see or alunah friend renew, your swap is set to 100. Way. So always like a gas price unit, you set your gas price 100 way, I’m gonna set my transaction to 101 way. And then if a miner sees my transaction sees your transaction, they’re just naively going to put mine first in the block, because they just saw transactions highest gas, the lowest gas to kind of incentivize people to pay the most gas. Another searcher might come in, though, now they see my transaction. And they might say, oh, Doug has been one to one. But I know this very profitable opportunity. So I’m happy to be 200 guests, I’ll pay even more gas. And I’ll still make money even after what I pay for gas, because I know this from earning opportunities very profitable, then they have to broadcast that to the mempool. Because how do they get miners, but that means now I can see it and I try to come in and auction over them, I’ll pay 300. And we go back and forth. It’s almost like a latency war in this peer to peer network where I’m trying to see transactions as fast as possible, because we’re both competing for the same opportunities, and the miners are kind of doing the same behavior. And there’s other searchers, I need to see their transaction. So I can bid as fast as possible, and get in line there without just bidding a crazy amount of gas and giving up all my profits. So that was kind of how the game worked in Formula One. And then flashlights came along and kind of changed the whole game.
Corey Hoffstein 33:32
We’ll get to that in a minute. It strikes me though, that like traditional high frequency trading, understanding toxic flow is really important here. Because let’s say I naively have a swap, you try to front run me. And another searcher then is able to front run you as well, the opportunity for them to profit is going to be on front running me and front running you. But if you then pull your transaction, the amount of gas fee that they were willing to pay to front run both of us is maybe very different than just front running me. And so this understanding of who’s a toxic transaction in the theory of mempool seems very parallel to the idea of understanding toxic flow in high frequency trading. I was wondering if you just comment on that.
Doug Colkitt 34:16
Yeah, directionally That’s right. I’d say the one wrinkle there is usually even if you’re a front runner, you’re not gonna get front run, because you’ve said your slippage, very tiny going back, right? It’s a smart contract chain. So I can say if someone has already moved, the price just got there before me just abort the whole transaction. I don’t want to trade. But like you said, you do pay a lot of gas and especially in gas floors, the price of gas you can get paid usually start scaling up to the total size of a profit opportunity. So there is a lot of toxic flow in the sense of you have to think about where am I in this pecking order? Am I the best searcher out there? And maybe if I am I’m just gonna go after every opportunity. Or maybe there are searchers I know who are better than me, but only do certain types of transactions. Maybe they don’t be able to certain types of tokens. There’s a whole issue around like, quote unquote salmonella tokens, which are like tokens that look like you can front run them. But actually, in the token itself, it locks the liquidity pool. So they’re basically a way to, like, take money away from searchers, there’s this whole risk curve, and you have to be aware of where you’re at on it. Because like you said, you don’t want to get on a gas board with someone who you know is going to be cheap, because you’re gonna end up paying a lot of money for gas without actually having any profit to show from it. And then you’ll lose money over time. So it’s kind of funny, like, one thing I would use is how many leading zeros off a given searcher. So if you ever see like a theory, metric, 0x 000000, whatever. Usually the fact that like, somebody went through the hassle, and like the cost to actually mine a bunch of zeros is an indicator that they’re pretty technically sophisticated. So if I saw someone else competing with me, who had 10 zeros or something, as leading digits could actually be a good heuristic. I don’t want to get bored with this person.
Corey Hoffstein 35:57
Sort of like, ironically, it’s like a vanity negative alpha signal. So you mentioned flash bots really quickly. You mentioned the way MeV worked in 2020, Flash bots came on the scene and dramatically changed the game. The game primarily probably being the game theory that’s being played among net searchers. Can you explain what flash bots is? And why the whole game theory of the way Neville is played changed? Pretty much overnight?
Doug Colkitt 36:29
Yeah. So before flat fonts, there was actually just a preface that there was this very brief period, like kind of like you said, maybe minor started saying, well, we should just do math directly ourselves, because no matter what the third party foragers do, we have ultimate control over the block. So maybe we should run our own MeV trading strategies, just internally put them in the block, and we know will always win if we control the block. That happened very briefly with his mining pool tried to get folic acid or something, they tried to run their own mess strategy and say, well, we should be getting some of this. So they had their own desk, they’re risking their own capital, basically. And the first week, they lost something on the order like a million dollars, and like a salmonella attack like a token that was like a fake Pokemon just designed to like, make them think they were front running, but take all the money they spent, then at that point, the miners were like, well, this is not worth it, there’s still a lot of risks, even if it seems like a good deal. We sin be involved with this directly. So the Flexbox came around, and flashlights basically set up this infrastructure parallel to the normal mempool, like we discussed, and with flat flops would do is they went to all the mining pools and they got them to agree, okay, connect to our infrastructure, people are going to send private transactions directly to you. So now it’s different, I can send a transaction to flush flux, no one else can see it, I’ll send a private transaction, I’ll set the price I want. And then the mining pool agrees that if it’s a flash flops transaction, it always builds at the top of the block. So if you’re trying to be a web searcher and going through the mempool, this breaks your game to someone else’s using Flash flops will always get the top of the block no matter what do. So this socks, or mempool, type searchers. And then flash flops have this infrastructure where they take away the mempool. Now transactions are sent privately directly to the miners themselves. What ended up happening is, like you said that changes to game theory, because before this auction where I could see your bed, you can see my bed, and we could try to respond to each other’s bids as soon as possible. You’re not incentivized pay anything more than except slightly more than your opponents pay, which means usually you don’t go all the way up to like the maximum gas value that we would pay. So therefore, the miners don’t end up extracting most of the value. Now with flip flops is a blind auction. So if it’s a blind auction, I don’t know what you’re going to pay. And you don’t know what I’m going to pay. But we each know what this opportunity is worth 10,000 gaps and whatever. So the incentive is pretty much we’re going to bid as close as possible to 10,000 gas because otherwise, you’re just not going to win someone else who’s just going to do that strategy, push it as close to the economic breakeven point, it drastically changed the value of crude oil from searchers were making a ton of money before because they were keeping most of the money. So now the miners most of value being generating things like 90%, or something pretty quickly with it.
Corey Hoffstein 39:16
At first glance, MeV seems like a very different high frequency game than, say market making on Turkish equities. At the very least, you’re operating like magnitudes of order, so lower on clock time. Curious more about what the commonalities are either in the technology stack, you need to run the types of alphas that you’re ultimately looking for, or even the game theory that occurs in these types of pursuits.
Doug Colkitt 39:46
I’d say even though the block times are longer, at least it’s mempool type game. The times you’re responding are actually faster because you’re playing this game in the mempool itself and you’re really only constrained by like global network banks. Those are still orders of magnitude slower milliseconds instead of microseconds. But you’re still trying to optimize this. Like, I want to respond as quick as possible, at least in that regime. It is different though, because there’s not one exchange. It’s not one server connected to an exchange, it’s this global network. Do you also play games where, okay, I’ll try to set up a bunch of different peers in a bunch of different geo locations. So I can always kind of respond first, no matter where the message originates from. So it is like networking games, but maybe not exactly the same type of networking games. And then like you said, from a strategy perspective, there’s definitely some aspect of like toxic flow, game theory type stuff. The biggest difference, though, is in high frequency trading, you’re really managing risk. There’s no way to do high frequency trading without having some inventory on some time, if a market making I’m going to have Phil’s be long or short, I have to manage those positions, I have to get in and out of it. Hopefully I do well, my Sharpe ratio is high enough, that doesn’t really matter that much. But I still have to manage it, I still have to take risk and hold the risk and hold inventory, a lot of these Meditite games, you’re in and out in a single block, or even a single transaction. So you’re not managing inventory, which is a whole different thing. Now, that being said, the standard pipe world, you might be managing inventory. And there have been really a few MEF players who have done really well at stat ARB. And a lot of players who have struggled to transition over to stat ARB. And the interesting thing about stat ARB is right, you might only do one leg of a transaction, hold the inventory, instead of getting in and out paying swap fees both ways, because that reduces what your costs are. And therefore you can bid on more opportunities, you can bid more, all kinds of stuff. So I think the big thing was a lot of people who got into Mevlut really understood blockchains, well, but didn’t have this traditional finance background of I know how to manage a portfolio, even if it’s a high frequency portfolio coming up, I know how to manage the portfolio, how to manage risk, and automatically it’s important, and the MEB players that got really good at that. And were able to do that. I think I’ve really taken a lot of market share from the more traditional atomic med sectors.
Corey Hoffstein 42:03
Philosophically, do you think nav is a benefit or a drain on the theory of ecosystem? For example, does it provide useful price discovery? Or is it purely creating congestion and its predatory?
Doug Colkitt 42:16
Run running I think is pretty much a straight negative. It’s bad user experience, people are getting worse Phil’s it doesn’t provide any price discovery because the price at the end is the same. Front running is a straight negative. It’s a fun game. But it doesn’t really help the ecosystem at all. backgrounding. And like atomic arbitrage, I think kind of helps the ecosystem because you’re keeping prices in line between different venues. And then similar stat are probably helps the ecosystem right I want the uniswap pool not to deviate too much from the binance or but at the end of the day. I think the strategies themselves are very simple. Most of the complexity is just winning the auction or getting to the top of the block. And like if you look at the amount of money that goes from ultimately liquidity providers into uniswap, pool to this types of traders, it’s really way out of line with like how complex it is to say, what’s the price of by Nance, okay? The price of by Nance is 50 basis points higher than it was 10 seconds ago, I should move the uniswap. So that’s like a pretty simple trade, probably the MEF players are getting way overpaid. Ultimately, profitable would be providers that will say the the one good thing though is it definitely pumps the value of cruel to the theory and the med players ultimately gas and pay fees to the validators and that increases the recurrence to staking rate, and so therefore, it burns more eath and all that stuff. So you’re an ease holder, technically, if you’re just holding it, you benefit from it.
Corey Hoffstein 43:43
It’s an interesting segue from there not to say that MeV ultimately was the catalyst for you wanting to launch Croc swap. But some of the things you’re talking about here, the inefficiencies in the way that things like the uniswap pool is designed for liquidity providers ultimately create an opportunity for MeV. And it’s something you’re trying to address with Croc swap. So let’s start with first, what is a decentralized exchange, maybe talk a little bit how they’ve evolved over time. And then we’ll take that into the core problem that you’re trying to solve with Croc swamp.
Doug Colkitt 44:19
So you centralized exchange is just anywhere you can swap one crypto asset or another. I guess you have NF T exchanges too. But we’ll just talk about tokens fungible tokens. So I have USD C and I want to use that to buy some Aetherium because I belong Aetherium or vice versa wanna sell Aetherium? So it’s single place on the blockchain smart contract that operates autonomously. There’s no venue, you don’t have to deposit your money at FTX and hope that FTX has your money at the end of the day. It’s still your key steal your crypto, go trade it you do a transaction tokens, go out of your wallet tokens come back to your wallet at the end of the day. So Oh, it’s very convenient in the sense that if I’m already holding assets on the blockchain, I just have to send a transaction through Metamask, or whatever my wallet is open and change it on deposit, I don’t have to hope that FTX still has my money or anything like that. That’s the purpose of it. Now, the question is really, how do you actually do that on the blockchain itself, computational environment are way, way, way more limited than what NASDAQ can do on like a big server. So the problem is, at least in Aetherium, or Aetherium, like blockchains, border books don’t really work, they’re just not efficient, we wanted to do one, the gas costs would get very high, very quick. And there’s all kinds of other practical issues. So order books don’t work. So most, almost all decentralized exchanges are what’s called an automated market maker. And it’s a pretty ingenious design. At its core, it’s pretty simple. There’s a pool of liquidity providers, they put in 50% of one asset and 50% of the other asset. And then anyone who wants to swap those deposits, some assets on one side, and then is allowed to take out assets on the other. And every swap slightly moves the price and who’s always trying to stay balanced, if the 50. So that’s how the mechanics work. The math is actually pretty simple. There are variations to it, notably, like concentrated liquidity where you can provide liquidity over a given price range, and which is more capital efficient, probably Aetherium, is not going to wick down to $10 in the next 10 minutes. So it’s kind of inefficient to be depositing tokens. So someone can go try to Etherium down to $10, it’s not going to work up to a million. So right, you can do concentrated liquidity. But the core idea is still simple, right? You just have a pool of liquidity providers are getting paid swap fees to always kind of be on the other side of the swaps. And hopefully the swap fees that they’re getting paid might be 0.3%, or whatever. It depends on the pool, the swap fees that they’re getting paid hopefully compensates them for being on the other side of which way people want to trade. So the biggest difference between that and an order book is an order book. Everyone’s competing to provide liquidity, which creates a very PvP dynamic. Let’s talk about all the new stuff. All these other things, some liquidity providers make way more money than other liquidity providers in order book. In an Amm. Everyone who’s providing liquidity is basically in it together. So it’s more cooperative form of providing liquidity. One of
Corey Hoffstein 47:23
the weirdest things about decentralized exchanges is this concept of providing liquidity at a given feet here. Can you explain how that works? Why it’s a potential problem for the long term setup of decentralized exchanges and how your proposed model at Croc swap works.
Doug Colkitt 47:42
The classic way that it works is each individual pool has a preset feed here that feed here’s defined when the pool is created. So traditionally, the install pools have been 0.3%. Meaning if I trade 100 USDC, I’m going to pay 30 cents to the pool. That’s the traditional model uniswap v3 now has different fee tiers. And the way that works is each for your tiers, technically, but totally separate pool. So there’s liquidity providers who are providing 0.05% liquidity providers who are providing 0.3%. And there’s a pool at 1%, which typically means like most of the time, people are going to route to the pool with the lowest fee tier until the price gets out of line enough to justify trading at the higher frontier. And that’s how so the features are just preset. And as a user, your challenge is actually selecting the right feet here. And it’s actually pretty difficult. What is the right level of compensation I should be getting for any given token pair. And typically like what you see is at least for volume, like outside, like all parents like STC point oh 5% is way too low. The liquidity providers and those pools are not being adequately compensated. But a lot of people just end up using it. Because it’s their inflicted, it’s kind of difficult to figure out. The system is pretty opaque. The system’s not responsive to markets, or pools defined technically people can move liquidity between features depending on market conditions in reality 95% Don’t when providing liquidity in an AMM, I just want to earn yield on it. I don’t want to have to monitor it 24/7 And then it creates a PvP dynamic. So if a lot of people are providing a point oh 5% And I’m providing a point 3% Now we’re competing with each other. So I think that kind of breaks one of the important models around Hmm, where increased PvP elements would mean that the best trading firms are going to win, which means that it’s not gonna be profitable to make a decent return as an ordinary person providing capital and on and on. So those are the biggest problems and yeah, just basically, it doesn’t change its status doesn’t respond to market conditions. Who knows 0.3% is right or not, and it’s probably the right level isn’t even the same from period to period or depending on who’s sending the orders. Certain order flow is a lot more toxic than others. So high frequency traders and Orderbook have all these tools to kind of making sure that they’re getting paid Need adequately for liquidity they’re providing they widen their spreads when markets are more volatile and narrow when less volatile, they’ll try to profile liquidity at certain times when they think it’s toxic or less toxic. Or they can go out the Robin Hood and said, I know your flows on toxic send it to me, I’ll charge you less than what you pay at NASDAQ. So basically, the idea what we’re trying to focus on is how can we bring better segmentation of order flow and time and using a dynamic fee model to see if we can fix some of the broken economics?
Corey Hoffstein 50:29
You know, let’s dive into that. Because it’s one of the most interesting ideas to me this notion of different fees based upon where the order flow is coming from charging a higher fee for toxic flow, for example, first, toxic flow is a phrase we’ve used a couple times, maybe you can define it just for folks who haven’t come across that phrase before. And then maybe you can give some practical examples of how charging a different fee would work in practice with Croc swap as it relates to sources on the Etherion blockchain.
Doug Colkitt 51:01
Lots of flow is basically just order flow that you don’t want to be on the other side of so if you have the choice to be a counterparty to all the trades that your grandmother is doing to rebalance her 401 K, where you have the choice to be the counterparty to all the trades that Citadel wants to execute, you’d obviously want to be a counterparty to your grandmother, instead of to Citadel so certain participants in the market, the fact that they’re trading carries a lot more information than other participants. And if I’m going to be a counterparty to them, which a liquidity provider or market maker is, I don’t need to get paid as much for somebody who is not necessarily a toxic Counterparty, I’m willing to charge them less for liquidity. So with that core idea, what you can see is dive into data, we’ve done a fair bit of this is across traders in uniswap, at least like on a five minute basis, how predictive is their trade over where the price is gonna go in the next five minutes? Almost everyone uniswap is non toxic 99% of people even like larger fund whales and stuff, they’re not moving the press, there’s a very small percent of traders, mostly MeV bots, are extremely toxic. Almost every trade they make uniswap is bad for the liquidity providers, they’re losing from that which most of what these traders are doing are just arbitrage in the price relative to by Nance or wherever or somewhere else price discoveries occurring. So our core idea is if someone sends an order, and there’s some credible signal that the order is non toxic, we can charge that order less than if somebody sends an order. And there’s really no credible signal, we don’t really have anything. So the most obvious example is, if you’ve ever used meta masks, you can swap tokens inside meta masks, they have a little swap button, which is very convenient. But they charge something like 0.8% fee or something on top of that. Sharpe trader is never ever going to send their order flow through the meta masks swap the router. So the fact that order flows coming from there is like almost certain that it’s non toxic, and you can charge less, almost down to zero, and still be profitable in expectation for liquidity providers. Another example might be if someone’s willing to send a swap, and they’re willing to wait 30 seconds until there’s swaps executed, they’re probably non toxic, because it’s usually nice, fast arbitrage. So you can charge them, let’s say if you send a swap, you let us hold it for 30 seconds, we can charge you a lot less. You can even go to the MEB traders themselves, you start profiling by address, and you can say hey, like I don’t need to know exactly what’s going on. I just know consistently over the past 30 days, whatever, all the order flow that originated from this router or this address, it hasn’t been toxic at all, it’s very easy to look at, we can look at the aggregate numbers, any individual trades hard to judge, but once you aggregate numbers to a large enough degree, you can be pretty certain. So that’s ultimately what our model is. Well two things. One is we’re looking at market conditions themselves. When markets are more volatile. Liquidity providers should be getting paid for liquidity because liquidity scares charge more for liquidity when demand is high supplies low. So that’s a time based model. And the other is an order flows based model and say there are different sources that are originating order flow, we’re going to get them a reputation over time. And people with better reputations can pay less swap things and then you become incentivized, I want to make sure my reputation is good, because that allows me to train cheaper. So the game theory kind of works out for that. And then there’s going to be some toxic flow, you have to make markets efficient. But if everyone else is kind of going in through these other channels, you can start charging the toxic flow more and more and fill the value accruals going to liquidity providers instead of the med searchers or the validators or flash flops or or anything else.
Corey Hoffstein 54:43
No one of the other major innovations that you’re working on with crock swap is the idea of vaults, helping solve potentially one of the big issues with decentralized exchanges, which is KYC problems, something that if we want institutional players to take defi seriously, this is a problem that needs to be solved. I’d love for you to sort of talk about your proposed solution here. And what else these vaults will enable.
Doug Colkitt 55:11
Yeah, exactly. Like you said, there’s a lot of players out there who will do maybe a ton of volume on centralized exchanges, and they might not do really do nothing on chain because the rules are ambiguous. But depending on if you’re fairly conservative, from a compliance standpoint, technically, when you interact with a liquidity pool, by some interpretation, you’re the counterparty to every single person providing liquidity in that liquidity pool. So obviously, if anyone from any address can go in and participate in liquidity pool, if I’m a compliance, Maxine, I might not necessarily want to interact with that liquidity pool, even if the fills are good. So our proposal for this is, we have something called quote unquote, permission pools. So we’ll always run with permissionless pools, so anyone can participate in the pool, permission pools or just a general purpose primitive. Basically, what happens is their same pool, same mechanics, but the ability to participate in that pool, or what you can do in that pool, or even the price you pay to participate in that pool, we’re going to parameters. So like, for example, where you can place concentrated liquidity, other things like that, those are outsourced to a general purpose Oracle, and then Oracle can be any smart contract. But for example, if you wanted to KYC only pool, one Oracle could be any address that interacts with this pool, or at least wants to provide liquidity in this pool. If an actress wants to provide liquidity in this pool, I have to make sure that it’s on this KYC whitelist, and has some credibility or whatever. And then someone goes to swap against a pool, they know, hey, I know, every liquidity provider in the school is KYC. So therefore, I know that I’m in compliance, that’s something really interesting you can do on again, there’s still the permissionless pool, so anyone can interact with those pools. And prices should stay fairly in line, especially when it’s relatively easy to trade between the two. The other thing is with permission pools, you can do a lot of really interesting things. Probably some of your listeners know about the Olympic style on like protocol on liquidity. And that was big last year. But the idea is Olympus out this protocol, basically owned 100% of the liquidity in their own pool. So yeah, it might make sense that the protocol itself would let people interact with the pool or pay different fees, or maybe only be able to buy at certain times of day or buy if they were validated members of community or they’d stick some other asset into it. So you can do a lot with the polls, sky’s the limit there.
Corey Hoffstein 57:36
Most people I know, in the high frequency trading space don’t tend to hold longer term views. I don’t want to put words in your mouth here. But I get the impression from speaking with you and reading some of your tweets that you are actually reasonably bullish on Aetherium. Over the long run. I’d love to press you on what your thesis is.
Doug Colkitt 57:57
Yes, you’re right, I am bullish on theorem and long term, my thesis is that the Etherium will become to the financial system, what the internet was to the telecom system. What I mean by that is, you know, back before the internet, you had all these telecom systems and interacting between one system to another was very haphazard, expensive, difficult. That’s why like, long distance phone call would cost whatever $5 a minute, or whatever crazy prices are, because you didn’t have a common interoperability layer. And I think what a theory really is, it’s, it’s a database that other databases can connect to, for example, if I go get a mortgage, the mortgage person has to check my credit score, they have to check that I’m employed, they have to check that I have enough assets in my bank account. And that’s all very hard for them to check because they don’t know if I’m gonna submit proof, and that’s up to date, and everything else. So they’re really just checking databases. Okay, your employee employed right now they’re checking on your employer, your bank account says you have X amount of money to make this downpayment, your credit score is this and it’s not out of date. So instead of it being a four week process, for example, it could be a 32nd process. You know, there’s gonna be corner cases, but in most cases, okay, check, check, check this all ones up, I can atomically confirm, everything’s aligned, and you’re not borrowing money and putting in your bank account of an app to go check your credit score again, to see if you have any debt outstanding. So I think it’s just as common interoperability layer reduces so many frictions in any sort of transactional system, but particularly finance. I don’t see how that doesn’t become everything just built. On top of that, just the same way if I have a network or anything, I just connect to the internet.
Corey Hoffstein 59:36
Last question for you. longtime listeners of my podcast know that I changed my cover art every season, much against the suggestions of my marketing friends. The inspiration this season is tarot cards just for a bit of fun. I’m having every guest pick a tarot card that will inspire the design of their particular cover. You chose the week Love fortune, I’d love to know why that card in particular spoke to you.
Doug Colkitt 1:00:04
Yeah, I think the description was it represents a turning, turning of systems or changing fortune or kind of change in general. And I guess one thing that I’m really optimistic is is that I think what we’re building is kind of a new model, not just for crypto or decentralized exchanges, but hopefully for financial systems in general. I think order books are great, but I think AML is ultimately if you can get the economics to work really will make markets a lot more efficient and kind of democratize and provide liquidity and therefore makes things more resilient.
Corey Hoffstein 1:00:39
Well, Doug, this has been fantastic. Thank you so much for joining me and best of luck with Croc swap. Thank you.