In this episode, Mike Townsend chats with Michael Safai, Managing Partner of Dexterity Capital, a quantitative trading firm focused on solving the hardest problems in finance. By combining brilliant people, world-class technology, and rich proprietary datasets, Dexterity has built an algorithmic execution engine that trades billions of dollars every day.
Dexterity Capital specializes in market-neutral strategies. That means they thrive regardless of whether the bulls or bears are winning. With offices in Seattle, San Francisco, and London they run their strategies 24/7 around the globe.
Mike: Thanks for tuning into Around The Coin. Today's interview is with Michael Safai, the managingpartner at Dexterity Capital. In this conversation, we talked about highfrequency trading, which is what Dexterity does. We talked about the influenceof TradFi and crypto. We talked about the international affairs andmacroeconomics and the growing tension between the US dollar and Bitcoin.
And all the implications that has. Wediscussed regulation, maybe some ideas around creating new regulation. And whatthe world looks like in the future. Michael is in a really unique position tobe operating in 2023 in high frequency trading. So we talked a bit about highfrequency trading, what the reasons that high frequency trading firms do well,what struggles there are influence on AI and many other ideas.
I hope you enjoy this conversation.Please do give us a share a thumbs up and without further ado, Michael.
Mike: All right, Michael.Thanks for hopping on. I'm excited to dive into some, some topics today withyou.
Maybe we can kick it off by talkingabout Dexterity Capital, I think talking about high frequency trading low, lowlatency, high frequency trading, and where, where that. Where that relevancy isin trading today as crypto is obviously less less active than it was in 2022.
I'm curious, just to set the table here,like h how many firms are active today relative to maybe a year, even two yearsago, and how do you sort of think about the market?
Michael: Yeah, it's changed alot in the past two years and there's this great, fantastic bull run in 2020.Just so many opportunities, so many things portrayed.
It was a lot of fun and during thattime, there were a few old school guys like us who've been doing this foryears, that are Crypton native. We know how things work in the crypto world.During 2021, we saw kind of two major types of entrants. Number one istraditional H F T shops that do stuff in trai.
So your towers, your jumps, yourcitadels. Those guys were all. And even jump that tower and specifically werekind of earlier to the game cuz they thought it was an exciting space. And thenyour second type that corrupt up in that window were kind of guys who were,newish to the game and just saw an opportunity cuz the market was so frothy.
It's kind of like, it's hard sometimesto say it wasn't free money at that stage of the market. There were so manypricing efficiencies, so many venues, so many exciting new products. There wereall kinds of way to succeed as a high frequency trader or even like a mediumfrequency trader in. So that was kind of the heyday was 2021.
But you know, 2022 Market cools, Tarablows up and we see a lot of those kind of new guys back away because unlikeus, they haven't seen bear market after bear market cycle after cycle. Theydidn't know what to do. So that's kind of one category. The other category,during that period, the traffic guys who had came in, they were still there andthey were still kicking even after the terror thing, doing a lot of volume.
It's presumably making a lot of profits.That'll change with ftx though. When FTX collapsed, people got really scared.That was a massive counterparty. People trusted it. They thought they were toobig to fail. They thought they had an American founder who was gonna beaccountable and reliable. Not some kind of something flung off offshore whereyou just don't have any recourse if something goes wrong.
When this. A lot of these traditionalfinance shops got freaked out and they backed away from the table. Not to saythey're gone they're still out there today, but they're not at scale. So Imight have, I've spoken to of course, lots of exchanges and one offshoreexchange I spoke to was speaking about one particular HFT shop that was traiand said, yeah, before ftx, they happily plunk down a hundred million dollarson our exchange and trust our custody.
These days, those guys are down to 10.They've really pulled. And then we had the Binance problem and that was evenscarier cuz you know, CFTC didn't quite name names, but they didn't make ithard to figure out who trading firms A, B and C were necessarily. So after thatwe've seen even more pullback, specifically from Binance, but also some otherplaces.
So it's a much smaller kind of cast ofcharacters in the H F T world in Crypto today.
Mike: And when you,
when you think about the, thedifferences between maybe folks like yourself who had been in crypto andthinking about things from a crypto first principles perspective and tradingcompared to people who came in during the 2001.
Rush on the Tradify side. Was there adifference in maybe the principles in which they operate or the technicalinfrastructure in which your side of things built up? Like what, what would bethe difference between people who sat down in front of a desk and do they pullup like, I imagine they're thinking, pull up a Bloomberg terminal and analyzegraphs.
I mean, is there a difference in how youactually do high frequency trading, or is it basically the same thing whenTradFi guys came in, they take the same playbook they've been doing with cornand commodities and whatever else, and then just apply it to crypto? Or islike, how does that work?
Michael: It's kind of a split,and we might talk about it, in terms of trades and strategies and infras.
And trades and strategies are, what'syour game plan? What are you gonna buy and sell very fast every day? And when Italked to people who came from TradFi, there were kind of two observations.One, they assumed they were gonna kick everybody's ass cuz they had all thegood infrastructure and they knew all the plays and they were gonna deploy thesame plays in crypto.
And the second thing there I would hearfrom them that was interesting was that, wait, this still works. This old tradethat hasn't worked in FX for a decade is still open. It's still on the table.Wow, this is easy. Let's go do it. So the strategy guys and the traders superexcited, saw lots of opportunity, saw things that hadn't worked in years, andTradify were suddenly available to them again, so they were happy.
But then you've got the infrastructureguys. Traders say, Hey, come and build an integration with Binance for me, orftx or Coinbase, whoever. They're like, cool. I've built so many integrationswith TRA five before I know exactly how to plug into exchanges. They typicallyuse something called a fixed connection, which is a standardized protocol thatyou use to feed your trades in and to listen to the exchange and action things.
So one of those infrastructure guys,when I sat down with the trader and said, cool, point me to the documentationfor the fix integration. I'll build it. And the trader says yeah, there's nofixed connection. Actually, they don't do that in crypto. This is all kind ofnew. Some people do have a fixed connection, but it wasn't standard the way itwas in tra.
They're like, okay, fine. I guess APIand web socket. Is that how it goes? Gimme the documentation for that. And theanswer was, yeah, sure. Here's the documentation. Some of it's in Korean, someof it's in Chinese, some of it's very old and hasn't been updated. Go build.And by the way, everything is gonna change every few days or few weeks.
Exchange is gonna push out new cha newchanges to how they integrate with their traders. And they may tell you thatthat's gonna happen. And they may not, they may retire the old documentation orthey might not, or they might not document it at all. So these guys werebuilding infrastructure who thought they were just gonna have a walk in the.
We're shocked because this is not nearlyas organized and simple or standardized as it is in the traditional world. Sothat was kind of shock number one to the infrastructure guys. Integrating withexchanges is really hard in crypto because it's a new cut, cutting edge spacethat changes everything all the time.
I think the second thing that was reallydifficult for them was around how, you know when you're getting data back, howdo you consume that data? They're used to consuming data in a standardized waythey get from the exchanges. Well, that's. And I think the third big thing forthem was all this infrastructure that they built for decades.
Like they have custom-built F BGAs. Theyhave dug holes through mountains to create a direct line of sight for themicrowave tower network. They've spent millions and millions building for TRAfive, doesn't matter, doesn't work in crypto. That big edge you thought youhad. It really isn't there. Crypto exchanges tend to be hosted in aws, justlike every kind of tech company, Amazon, whoever.
They're used to get plugging into aco-location facility where the exchange has a physical machine and they puttheir box as close as they physically can by paying a big fee to get next tothem, and then optimizing how short their cable is. So the latency isminimized. All that expertise, pretty much worthless in crypto, doesn't helpyou.
So short. Long story short thought itwas gonna be easy. They're gonna come kick ass cause they've been doing thisfor such a long time. Totally different rules of the game. And that was.
On that third point there theinfrastructure, the, in 2021, when you know everything is growing and there's alot of inefficiency in the market, does, does the tightness of the code, theefficiency of the code, the frequency of the trades, does that matter as muchif there's so much inefficiency or is high frequency trading more relevant whenthe market is tighter and more efficient?
I guess less it would be, it makes senseto me that it'd be when the market's more volatile, high frequency trading ismore relevant. But like what, what are the conditions for which high frequencytrading would be most relevant?
Michael: Yeah, I think, Ithink, here, here's the thing, that's another big split between tradify andcrypto.
If you're an HFT shop in TradFi. Youthink of things in nanoseconds or microseconds. So 10 to the negative six, 10to the negative nine. Very small numbers. It turns out that in kind of cryptoworld, since you're all gated by this AWS infrastructure, or Alibaba Cloud orsome other cloud provider, the best you can do is gonna be on the order ofmilliseconds.
So a thousand x slower than they're usedto. Wow. So yeah, you can go optimize your code, you can go get really low. Butwhen you do that, all you're saving is microseconds or mics as welo call them,right? Well, that doesn't matter because you are on the millisecond scale now.So even if you improve 20 mics, which is a big deal in tradify, does not matterin crypto, you've got the speed bump, it's called cloud infrastructure, andit's gonna limit you to that millisecond area.
So I think that's kind of the big changethat they had to go through In terms of maybe more directly to yourquestion.
Mike: What are theconditions for maximum relevancy for high frequency trading?
Like when, when are you sitting backbeing like, now's a great time to be a high frequency trader in crypto.
Michael: So traders, tradetraders like us trade on volatility and volume. So moments when there's a lotof trading happening is definitely a big plus. And we like that and we thrive.Then when things are quiet, not so great.
You still make. The way we think aboutit is, I'm gonna do X volume and all my volume. I'm gonna make five bps, orfive 100th of 1%. So a really small number on that volume. So I take five BPSand multiply it by X volume. That's how much money I make that day. So ofcourse if volume goes down, you make less money.
That sucks, but you still make money,which is great if you're a Shopify gars especially. That's market neutral. Wedon't care if prices go up or down. We feed on volatility. So high volume timesare the. But there's a caveat because when all these exchanges are new andthey're all cutting edge, when there's a lot of activity, things go wrong.
If you have an exchange where you canusually send orders within 20 milliseconds, for example, during a quiet time,it might spike to like 300 milliseconds. Like FTX was actually particularly badat this. During periods of stress, your roundtrip latency would go up 10, 15 xand that changes things and you have to be ready to catch those lagging datafeeds.
How, how many, if you had to guess, howmany people are doing this full-time, how many people are sitting in yourposition as a full-time crypto, high frequency trader?
Michael: Not a lot. I mean, I'mgonna say probably the people who are really good at it on the order ofthousands in the planet. Oh, okay.
It's a really hard thing. It's reallyspecialized. And you've got to. It's, it's information that's not publiclyavailable. You can't go Google, how do I high frequency trade and get a goodguide? You won't. It's all knowledge you have to get by blood, sweat, andtears. So the people who can do it have to be incredibly smart.
I'm just, for the record, I personallyam not a trader. I'm the dumbest guy on the team. Proud of that actually. Butgenerally my guys that I've met, they should be curing cancer. That's what theyshould be doing. They're way smarter than your average. So there's just veryfew people who qualify for this intellectually, and then you have to get intothat kind of knowledge of how you actually execute.
You can have the raw p horsepower ofintelligence and money and hardware, but you have to figure out a way to crackwhatever the super clandestine secret crypto system is and compete with thebest in the world. So really single thousands is my guess of how many peoplework on this full-time.
Mike: And, and when youthink about, this is kind of a, a good transition,
when you think about the upstreambenefits of society that frequency, high frequency traders provide, or maybeeven just crypto traders more broadly, do you think of it as providingefficient prices across different sources, maybe exchanges or yeah, ba basicallyall the different exchanges out there so that when someone comes into theecosystem, it makes sense.
Bitcoin's. 20,000 over here and 24,000over here. There's like tight bandwidths and that efficiency and pricing acrossdifferent locations makes the whole system more stable. Is is that how youwould describe it or would you describe it differently?
Michael: No, I, I think that'smostly right. I'm, I'm not gonna tell you, we're saving the world.
We're not we're making money off of veryminute dip price differences across exchanges. But if you're a retail. Thepresence of people like us who do unify prices, who make sure have for theirown benefit, of course, that the price of Bitcoin, for example, is the same onCoinbase finance, Kraken, everywhere else.
That's actually a good thing because asa retail trader, you're on equal footing with whatever platform you can tradeon in the world. And crypto is global. And it would not be very cool if, Hey,I'm a, I'm trading in Germany and I should get a very different price thansomeone trading in America. Part of what we do helps tie prices.
For the most part, there are some placeswhere there are premiums, like you might have heard, like Korean premiums.Sometimes even when you convert one to dollar, you'll have a higher price forBitcoin on Korean exchanges or Nigeria's. Another good example, and there's afew reasons this happens, sometimes it's because in the case of Korea, there'scurrency controls.
They're very strict about who can tradein Korea. And you look at Nigeria, you've got kind of a hyperinflationsituation there and. You have to be very careful with who you let trade inNigeria. So there's some reasons prices diverge, but in the places where therearen't currency controls or the government's not really meddling with thespecifics of that, it's generally a good thing for retail that we're all on thesame stage.
What, how about, let's talk aboutexchanges more broadly. I'm, I'm, I think this is just a super relevant timefor the, the way that governments and exchanges. Relate to each other. Thereseems to be just an obviously growing tension, at least in the US between the,the relatively few exchanges out there, which I, I would love to see dozens.
I would love to see dozens of thrivingexchanges, or at least different ways that you could exchange u s D to BTC andget into the crypto ecosystem. But there's, I mean, how many, how many arethere? Right? There's like Coinbase, Kraken, The ftx, rest and Peace Binance,like you can count on two hands.
I mean one, one hand, two hands. Likehow, and is that, is that, is that a target by governments? How, how overtly doyou think the United States and, and other leading Western countries viewBitcoin and cryptocurrency as a target, as something they're maybe behindclosed doors that they're really trying to pressure out of existence?
Is that is how, how do you think aboutthat tension?
Michael: Well, it's kind offunny. Initially we thought of this, a few years back, there's all these exchanges.Consolidation must happen one day, right? If you look at the NCI or nasdaq,there's kind of only so many big exchanges out there in the world, and that'swhat we thought was gonna happen.
We thought exchanges would buy eachother up. Didn't really happen that way. Acquisitions are actually kind of farand few between. For exchanges. Sometimes they'll buy something cuz it's insome region that's underserved. They don't have the customers there, so they'llgo buy an exchange in Thailand or something.
But for the most, They don't reallyconsolidate. So, that was our expectation. What caught us a little off guard isthat what really happened was failures and regulation is causing consolidation,and that's for two reasons. One, people are now really careful withcounterparty risk. They're not gonna go deploy a million dollars, which is asmall number for a lot of traders have, have kind of scaled to a littleexchange somewhere in the world because they may never see that million dollarsagain.
So that was kind of the big change. Thesecond thing, as you mentioned, is regulation. And I, I don't have any personalinsight into what's happening with the CFTC and S E c I wish I had an insideline. I wish they'd called me first, but they don't. But what you can kind ofput the pieces together, the US feels threatened by crypto.
And the u s d, the dollar is is kindathe world's currency to a certain degree and it's very important for theAmerican government to have hands on that they want to grab, like be in controlof where dollars. And when they saw what was happening with stablecoin andexchanges popping up everywhere, they really kind of didn't like that.
That goes against the narrative ofcontrolling your currency. When we saw FTX fail, that was almost a goldenopportunity for people within the government who wanted to stop this fromhappening. Cuz suddenly it was a clear threat to retail investors. Americanslost money on FTX ftx. So when they saw this, it became an easier pitch tosell.
And my belief is at this. Yeah, us wouldbe a lot happier. Crypto just.
Mike: Yeah, and you can'tforget Celsius cuz my understanding is FTX affected international. There wastwo different accounts, two different, almost like company operating manualswhere the US version of FTX, I think is solvent, although it's, it's locked.
I don't know if people have actuallyreceived any money, but the Celsius was a American operating company. Verylarge not an exchange. They were more a what would you call that? Like class ofa lender.
Michael: Lender. Right. Generalyield for people to them. And they would lend out to people like us.
Mike: Right, exactly. Likecentralized crypto lenders, which in hindsight, hindsight, 2020, it's somethingto be very skeptical of. Do you feel that government regulation is a. Is isobviously a positive thing. I mean, just devil's advocate here. You could havean ecosystem that develops where a few exchanges break the rules.
They do things, they screw overcustomers, everyone learns from their mistakes, and then a slew of companiescome in to effectively provide free market assessments or ratings of differentcrypto projects. In fact, I know that's what you guys originally were trying todo is provide different ratings for crypto projects out there.
Is that not a. Ideal pathway to have themarket provide a rating system and an auditing system for exchanges instead ofthe government. I mean, h how do you look at the bet, pros and cons ofgovernment versus free market?
Michael: Yeah, I think, youknow what we call them in traditional finances like sro, self-regulatoryorganization like finra, technically not a government entity.
It's actually something that's run bythe industry to make sure there's all fair play and the benefit to theindustry. If they regulate themselves and they do it well, the government's notgonna come in and change things. Crypto was harder because it was kind of ascattered set of players who were making decisions.
They were competing with each other.They didn't know what jurisdiction they were gonna be in anyways. They kind ofhopped around jurisdictions like Binance has moved a number of times, I think.So for them, like getting all together and creating a self-regulatoryorganization just never happened.
People tried, but it didn't. So I think,in an ideal world, that's what should happen, be happening. And I think, look,I want regulation, and I think a lot of exchanges want regulation. Coinbasewent and talked to the S E C and the CFTC many, many times in the years, triedto work with them D I D X, same deal.
They were very close to kind of the CFTCand what they were building to make sure that they don't kind of aggravatethem. Unfortunately, they never got clear guidelines. They never got a clear storyon how to operate things from the. So they kind of had to guess and they wereguessing on frameworks that are very old.
Simple question of what's a security?You go back to the Howie tests, which is a, a court ruling in the US and try toapply that doesn't really work great for something that's decentralized. One ofthe core things besides with security is whether it's managed centrally. Sodoes decentralization adequately get rid of that and therefore say you are nota security under the Howie test.
They couldn't get an answer from theregulators. Hmm. And I think part of that is it's a really hard problem.Crypto's weird. It's new. You've got a bunch of kind stodgy old folks who arerunning these organizations. They're not gonna understand it. They're not gonnareally care about it as tiny compared to what's on their radar.
And then second, back to my earlierpoint, it was a threat to the US dollar hegemony. That's really important toAmerica politically. I think that's really what happened here. People wantedregulation. We didn't get it. And now the s e C and CFTC are kind of takingheads.
So where does this go, right? Like doyou see, do you see this pointing towards like greater than 50% chance? There'sjust a straight up. Like, I don't wanna use the word war because it, it kindof, connotates like shooting missiles and, and that sort of thing. But this is,this is more, I don't think there'll be any missiles flying, but there'll be alot of lawyers involved.
A lot of threats. Maybe even like a, aphysical threats, like the government has guns, like cryptocurrency, exchangesdon't. So therefore, like what you either get out of the country or. Like wewe're just gonna come knock on your door and shut you down. I mean, do, do yousee any way, or should the community of people in crypto be doing anything Now,I, for instance, I saw Texas and Florida, Wyoming has some Dow laws, but Texasand Florida were pretty recently explicit in how they're going to respectpeople's property rights for Bitcoin.
I would imagine that that's going tolean towards more conservative red states. But like, sh who are the shouldlike, I don't know, leaders in crypto, be trying to advocate for any sort ofpreparation today before, I don't wanna say before it's too late, but I, Idon't know how else to say that.
Michael: No. And I, I think,they've been advocating for clarity on how to operate and they're just notgetting it. So what's going to happen? Nobody knows. The way we're headingright now is that, look, crypto's not going away. You can't get rid of itenough. People love it for whatever reason they do, and there's lots of reasonsto love it.
What will likely happen is that it'll betwo systems you'll have everywhere else offshore. So you'll have what'shappening in Asia and Europe, be under kind of a different umbrella. And thenwhat's happening in the states and the state seems to be kind of wantingchecking out of crypto. They just don't want 'em to be part.
The financial ecosystem. So if I had tovet right now and I'm usually wrong that's where I think we're going. Crypto'sgoing to be a non-US enterprise for the most part, but I hope I'm wrong.Yeah.
Mike: Well, okay, so I, no,I, I think you could quite simply observe the incentives, like United States isa massive incentive to keep the United States as a global reserve currency.
The number I heard was 86% of US dollarsin circulation are held by people outside the United States. And so when thegovernment prints money, gives it to their own citizens it's basically a tax onthe rest of the world. But it benefits the. A asymmetrically so us has anincentive to keep it there.
If people just have a free exit, doorsopen to Bitcoin, then that presents a threat and so it would make perfectsense. This is kind of B stream of Austin's whole pitch, is that people aregonna be flooding for the exits faster and faster. So get out now, and he has afamous million dollar bet on this.
I certainly can see all. Large chestpieces aligning in that direct. It just seems like such a uncertain settledstate, right? Like are small countries like Palau and the British VirginIslands, Bahamas, are these places just gonna be exploding with innovation?People are moving there, they're creating large exchanges, and then the futureof the internet and decentralized cryptocurrency.
Is all happening there. The obviousconsideration is that, like I, I think about the psychedelic ban on psychedelicmedicines back in the seventies. The United States basically said, no, we'renot allowing this. And then they used their influence across the world topressure other smaller countries to take on the same regulation, and thereforeit was pretty much stopped worldwide, at least in the West.
And so the obvious question is what isthe range or the influence that the US. In pushing their anti crypto policiesto smaller countries, like could they effectively stifle any innovation by, bypressuring financial regulations in other smaller countries. That to me seemslike where the real attention is.
Cause everything prior to that point, Idon't see how else it could go. It seems to be the pattern. Ray Dalio's famousfor proclaiming the rise and fall of civilizations. They seem to have thisseries. Properties that happen after they start printing money and go into highdebt to GDP ratios later in their life cycles.
It's like the United States is like, oh,there's, there's where we are in the, in the, on the cycle. And you can sort ofsee things happening as they have been observed in other. Civilizations. Butthe, the, to me it's like the, is that, is that going to happen? Are peoplegoing to literally move? Like, would you move to say London, say England cameout and said, Hey, we're just gonna ban cryptocurrency.
Is it, would you move, would you, wouldyou guys move the company to a smaller country? And maybe do you see otherpeople in your friend circle like being willing to leave countries to continuetheir work? Because that's seems
Michael: I won't speak about.Yeah, I think you're right that there's a thread of like a brain drain.
Mm-hmm. Or what's kind of the cuttingedge of FinTech. Right. So it's a crypto brain drain. And I, I won't speak formyself or Dexterity because there's lots of reasons to move businesses want,but I like my pubs. I like my house. I'm probably, I know, I know. That's thething we all do. But what you reposit it has already happened.
SVF famously moved to Hong. Yeah, I knowcountless Americans who left the States and went to Hong Kong, Singapore Panamawas a hotspot for a while, zoo and Switzerland. They went to places wherejurisdictions were friendlier to what they were trying to achieve. So that'salready been happening and I think anyone who's going to want to do crypto hasto consider that.
If you're a founder and you really, thefocus of your life is that, It might be a really good decision to get outta thestates at this stage. I love the states. It's a beautiful country. It's just,there's no place like it on earth, but at the end of the day, you wanna build abusiness, you have to make that tough choice.
So, we've seen it happen. I think it'llcontinue to happen. I think your earlier point that the US has a lot of kind ofclout throughout the world and can enforce regulations internationally isdefinitely true. And that's not necessarily gonna change anytime. Many of myfriends out here in Europe complain that they have nothing to do with thestates.
Maybe they've never been there. They goto open a normal checking account at their local bank in Austria or whereverthey are. They always get asked, are you an American citizen? And Austria's notasking that for their benefit. It's they're asking it for the American benefitwho has their kind of fingers in every banking system around the world.
So they'll definitely try to use thatleverage. But look, look at the world today. It's not a coincidence that thiscrypto squeeze is happening. We're seeing this kind of access developed betweenRussia and China and parts of the Middle East and this kind of global, it's nota full on Cold War, but it's getting there.
It's happening. And a big piece of thatis what happens with finance. So will that clout matter if China formallybecomes an enemy? I don't actually know because what leverage do they have? Andthat's what we're gonna have to wait and see, depending on how the macrosituation evolves.
Hard pivot to ai. Do you see AIpresenting a, like, I don't wanna label it as good or bad, but a maybe moreefficient or less efficient influence on. Trading, like when you're sittingthere, I, I presume that you guys have a lot of code running in the back end.You're, you're probably designing that code with some concepts in mind.
Some goals in mind is AI, from yourvantage point, something that's going to drastically improve the efficiency of,call it like back office trading, high frequency or otherwise, some way allalgorithmic trading, I imagine. Probably is going to be highly influenced byai. Is it already today? Like where do you see that influence happening?
Michael: I think definitely inthose back office functions and a lot of the more mundane things. This is amassive boom to have AI to cut out a lot of the things that you have to do thehard way, just using human minds. But there's a problem here. There's a lot of,I won't name any names, but there's a lot of different places out there whereif you're just an average retail trader, you give them your API key for say,bin, and they say, we'll run automated trades for you.
And that's sounds great. If you're justa retail guy who's knows nothing about H F T, problem is you're using it. Soare another 10,000 people. They're all using effectively the same bot. And thenthe commercial guys like us who our job is to identify what those bots aredoing and try to beat them. And we do that through a lot of.
So my question to you is, yes, AI couldchange a lot of how people trade, but if everyone's relying on chat, GBT fouror whatever, right? They're all relying on the same AI tech. They're just gonnabe like equally matched. It won't change anything. And the difference is if youcall AI 99% of the work, that last 1%, that's human is gonna decide the winnersand it, it already does to a certain degree.
I have lots of smart people working forme, all the other crypto native shops. Certainly the big tradify shops do, andwe're all butting heads trying to get as much as we can against really smartopponents. So look, we spent years building our infrastructure. We spent yearsdeveloping strategies, but they constantly need updating and human kind ofintervention to make them better and to get ready for what the opposition isdoing.
Mike: Yeah. It almost feelsto me like, you have a a, a line of a hundred. Athletes on a racetrack goingaround in a oval, and all of a sudden the racetrack just went from dirt torubber. Everyone is much faster, but the difference between who comes in firstand second and third might be a matter of inches.
And so the, the last 1% really is whereall the, the gains or losses are made. And AI is just like the infrastructuralboost to everybody.
Michael: I think that's ahundred percent right. I, I watch Jeff Juan, and you can see two drivers likeRed Bull, hundred milling the same car. But the difference between extra stopand Checo is they're two different people.
Yeah. And that's what decides thewinner.
Mike: Yeah. Superinteresting. How,
how do you feel about AI more broadlyin, in either crypto or even more broadly in the world? Do, do you see it assomething that's going to be a short term, massive influence to economicproduction, do you think the. The perception on that is over-inflated.
Maybe the actual tactical integration ofAI into people's daily lives is not as quick and seamless as people make it outto. It's certainly the ability to make, write blog articles, prepare, legaldocs. Like those things seem to just be happening, but whether or not it, likeit actually connects and integrates into complex worlds that have multipleinputs across different software systems.
I, I'm, I'm less convinced. I don't, Idon't know. What's your take on it?
Michael: Well, I think, Ithink, I used to be a lawyer and man, I wish we had ChatGPT four when I was alawyer. Cause I spent a lot of time writing stuff that I probably could havedone in an hour with this. At the end of the day, I still have to go argue withthe judge.
Mm-hmm. And I know whether the judge,had a bad warning or not. I know what the judge, like if he went fishing andhe's happy or not, or his wife left to me yesterday and he's pissed. And oftenthat decides a case, believe it or not. So if I, even if I had AI in my legalwork it's not the whole solution.
The good lawyers will still distinguishthemselves from the poor ones. Mm-hmm. So that little, that last human bit, Idon't think it's gonna be automated soon enough to really matter right now, butit definitely changes the game. And look, this isn't the first time you'veexperienced this as a civilization.
If you think of a good example ofmedicine as a radiologist who looks at x-rays and MRIs, all. And back inprobably the seventies, this was an incredibly lucrative business to be in ifyou want to be a doctor Now, however, it's fundamentally changed because ittakes them two seconds to send that to somebody in India who has a lower costof living, but is also a good doctor, and they can outsource the reviews of thex-rays and the MRIs, and then send back the results.
And then a doctor in America does thefinal step of delivering the results or interpreting them and discussing themwith the. And that was a big shift in that business. And we've seen it in Techtwo with engineers worldwide working on a project that's, for example in the usthat's happened. This is kind of the next step where we eliminate certain stepsand it's not so much outsourcing it to people, but outsourcing it to machinery.
So we'll see what that, what that endsup. But I can't imagine a world in near term where AI is in a big piece of it.Mm-hmm.
How sophisticated are people right now,let's say traders on their trading techniques? Like, I'm trying to get a sensefor. they taking into account data from APIs from some satellite company outthere and then crunching where trucks are moving and trying to figure out,like, are, are they, are they at that level where they're incorporating allsorts of data sets?
Like that in, in all sorts of unrelatedand uncorrelated sources, or do you feel it's more primitive than that? Thatpeople are really trying to just analyze graphs of Bitcoin versus Ethereumversus all the other cryptocurrencies and they're purely looking at it withinthe confines of crypto. Like, gimme a sense for the sophistication of humanbeings behind the scenes in training.
Michael: So mean if we talkabout what you described as like, let's call it third party data. Mm-hmm. Causewhen I get data from finance, for example, I listen to all what's happening onexchange, let's, you can call that second or first party depending on how youthink of it. There's all these other data sets, like you said, what the weatheris, where the trucks are, what are people tweeting?
If you look at traditional finance,people have been consuming that and researching that for years and years andyears. It's bread and butter for many hedge funds. Everyone knows that.Crypto's definitely lagging in that department just because it's still new andthere's so much low hanging fruit.
The markets remain so inefficient.There's so many venues that you can make profits without spending, a hundredthousand dollars a month for some third party data source. So people areprimarily in that stage of the market, and as it evolves, people will movetowards integrating those additional signals that are derived from third partydata.
But even still, I've known crypto nativefunds that for years, their whole play is, we scrape Twitter, we do it fasterthan anybody else. We know how to interpret emojis really well, cuz emojis arefar more discreet than words. And based on the emoji and the pair and thetweet, whatever Coin it is, I don't know, like flair or or salon or whateverthey can make decisions about where they think the price is going.
So it's being done. But it's not nearlyas sophisticated as directly.
Mike: So, so would you agreethat the, the difference between the 98 and 98 99 percentile performers ontrading might, might be not the ability to assess the data on Twitter andscrape it, but the, the actual ability to. Gather that data, whether it's froman api, from satellites, or scraping off some website somewhere, or Twitter,like is it really just going to be about a race to collect data and theneveryone's using the same tools like AI tools to analyze that data?
Michael: I think that'sfeasible. But I think, trading shops are a little funny. They're not likeWalmart where they sell everything. They don't run every kind of strategy.Mm-hmm. Right. Often shops zero in on a strategy. They're really good. And thenthey might spend the next decade improving it and fighting off people who tryto steal their lunch.
So that's a big part of, especially thesmaller shops. A lot of the kryptonite shop shops like ours, we have our breadand butter and that's what we focus on. And I don't care how much third partydata you give me, it might not help that. So it's hard to say when you thinkabout who's better or who's doing a better job of getting stuff.
It's hard to say if that's really whatthe key is. I think it'll be more important as we go. When someone hears, yeah,there's a signal in the data about how much ice cream is sold that day and howit correlates to Bitcoin. Well that's interesting. Everyone's gonna grab thatand it's gonna become part of the baseline of how you trade.
That last little bit comes down to whatyour bread and butter is and what you know that nobody else knows.
Mike: I, and is it I,
is it about. Our secret's valuable intrading. Like if you were to just tell me now, hey, here's exactly our breadand butter. Here's, here's kind of how we think about it. Here's what we know,here's what we don't know.
Here's what we're experimenting with.Here's our approach. If you were to spell all that out, does that, is that a,is that commonplace in trading? Is that like, oh no, I would never do thatbecause that's our valuable ip? Or is it more like, the execution of it is thevaluable thing? Yeah, Mike, here's how we do everything.
No big deal.
Michael: Well, my execution hassecrets too. So secrets are kind of the crux of trading at the end of the day.Yeah, it's what you know that nobody else knows. A a really simple example, onetime we were trading on an exchange and we found this API that wasn'tdocumented and we plugged into it and we're like, wow, this is 10 millisecondsfaster than what everybody else is using.
Let's keep using that. And our, our pand l went kind of through the roof and then it was quite funny cause maybe amonth or two later something. That like API didn't give us the right data. Wecalled the exchange and said, Hey, this, this endpoint broke. Can you tell uswhat's going on? They're like, is it broken?
We don't know. I'm like, well, haven'tother people told you about it? They're like, no, no one else has told us itbroke. And that's when we knew it was a secret.
Mike: Ah, interesting.That's a good example actually. That's a really good example. Yeah.Interesting.
Michael: Yeah. So that, thatmade a bottom line difference to us.
So, so secrets are like at, at the coreof trading shops and you've seen trading shops are very aggressive. NDA is inconfidentiality agreements and enforcing them. That's really the key. Andthere's definitely other creating shops who want to steal those secrets.Mm-hmm. And who will interview your people and try to get them to spillsomething, or may even hire them and try to get them to bring code with them.
That's certainly, wow. Wow. That's superinteresting. So it's, it's not, it's like kind of, it's CIA ish today, if youknow what I mean. Yeah. And we're very, we're very careful about what we allowpeople to know within the company because, not cause we don't trust them. Younever know when something might go outdoors.
Mike: That's fascinating.And so it's, it's, yeah, right. Integrating with the API is 1% of the work.Actually finding the API and realizing this is a valuable insight is, is wherethe vast majority of the value is. Not to say that actually building it andintegrating the data isn't hard or isn't itself something that everyone coulddo. And how would you pay?
Michael: Yeah, I think 99% oflike 99% of winning the lottery is buying a ticket, but at the end of the day,that last 1% decides whether you've got a hundred million or not. Totally. Samewith trading. You have to spend all that money to get in there and all thatresearch and expertise. But that last little bit, That decides whether you winor
And, and how, how would you describe,are there different themes to these I don't wanna call 'em secrets, but likestrategies, like you mentioned Twitter. Is there like a, do you think of thisas, okay, if you're to paint the, the map of different trading firms out there,do you have like the social guys, the guys trying to get indicators off of whatpeople are talking about socially in the internet?
And then you have the, is there, isthere some kind of high level macro map that you can draw?
Michael: Everyone's sosecretive in clandestine, it's hard to point to who's doing what. And it'soften hard to kind of discern it. And, when I talk to other traders, we don'tsay much to each other about the trading.
Not because we'd spill a secret, butsomething we might say might trigger something in the listener's mind thatsays, oh, there's a connection there. We should go investigate that and go findout what the secrets are. So I think, I think that's what it comes down to isevery little secret can. Be what decides your fate and fortune.
So why do podcasts, why, why go outhere? You guys aren't raising money. You are. I would love to talk a little bitabout prop trading, not taking money from investors, but before we do that,what, what do you operate the business? Like, what's the strategy of thebusiness? Why do podcasts, why have an externally facing brand?
Where do you guys want to
Michael: Yeah, I think, I thinkyou know, why we actually talk to people, which most prop shops don't do isbecause I think part of the maturation of this market space needs to be, thereneeds to be some knowledge to it. If everything's behind closed doors,regulators don't know what to do.
People don't know what to do. Peopledon't know what's important. And counterparties think they can get away with anythingcuz nobody's talking. Right. And that's not what I want this industry to be.Trap is not exactly my jam. There's a lot of things they've done right. Andpart of it is having a higher degree of transparency in the institutions in trathan we have in crypto.
That's really important. So I thinkthat's really what motivates us to go talk to people and tell 'em at least alittle bit about what's happening. I don't really tell you anything too secretbut we want, we want people to know that this stuff is happening so it can beaddressed, so it can be intelligently assessed by people who are going toregulate it or be accountable.
And if you could wave a magic wand andhave any political or societal change as it relates to policies, regulation, isthere something that you're striving for or would like to see in theworld?
Michael: I think the mostimportant thing is clarity. As I said earlier, a lot of people, a lot of regulatorsaren't equipped to understand what's happening in crypto.
They try to regulate it and they say,well, this isn't gonna work cause of xyz. So they just don't even bother. Orthey do regulate it and it's so heavy handed because they don't understand thenuance. And part of that's because nobody's talking. So what I really want isgovernments, regulators to come out and say, you can do this.
You can't do that. And for them to do itintelligently and with the consultation of people who understand thespace.
Mike: Have you ever seen, I,I've always thought this, cause I've heard that explanation on other, what I'veinterviewed other founders on other crypto projects and, and I've oftenthought, why not?
Why not just do it? So what if you and,four or five other smart traders out there got together and say, let's justwrite a Google Doc. Like if we're in, if we're leading the s e C. Or whatever,this is what we think it should be. And just have comments as to why you havethis specific policy. Like, okay, if we want greater controls, we would dothis.
If we want less, we would do this andkind of almost from a legal approach, like write the doc, write comments, showwhy you made a specific wording of a specific policy the way you did, and justopen source it like open letter to the s e c, please consider this. Cause Iimagine they're sitting there. None of them have experience in crypto.
None of them, they're maybe evenlistening to this podcast or other podcasts trying to parse out what's actuallyhappening. Maybe they're opening up their own account, trying to figure it out.I mean, they might be complete newbies, and I often hear smart people likeyourself saying, oh, we want, we want great policy.
If you were to just like, make it easyfor them to buy, make it easy for them to say yes, like, here's the dock. Opensource. Don't even gimme credit. Just take it, put your own name on it. Call ityour own bill. Like, have you seen that done? Why not do it? Is that a stupididea? Like what, what, what are your thoughts on that?
Michael: I think that peoplewho have tried to do it, like if you look at, at the Blockchain association indc, definitely they're on the hill. They're working with people all day long,trying to kind of help people understand what's happening here, why it'simportant to the world, and what the nuances are, so it doesn't get regulatedin a kind of negative.
I wish I could do it. Unfortunately,it's, it's time is what it comes down to. At the end of the day, am I gonna gospend my hours doing that or trying to boost p and l? And it's a toughdecision. Yeah, because it's really what I'd like to see. And this is part ofthe problem. The incentives are all wrong here.
And then again, earlier, as I said,people like Coinbase do idx. They have tried to do this and we've seen wherethey've gotten, they're kind of hitting a wall. So if regulators like want totalk come. But they just haven't been re receptive to that yet.
Mike: So hy hypothetically,Two, two hours a week.
I just, I feel, I mean, not to push youin any direction, but given your background as a lawyer spot, man, I love it. Ifeel like you have, you have such a unique credibility in that you were alawyer and you're running a high frequency trading firm and you're successfulat it, and you've been doing it for a while, and you have, you have other smartpe like you're very, you're not sketchy, like not individually, but you're all,you're not sketchy from.
From a, like a outsider's viewpoint. Andso if you were to get together and say, okay, we got 10 other leaders inrespectable firms. We use ai, I don't know, use something to draft it up andyou guys go through, spend a couple hours editing it, get out like an MVP tothe world just to just to do the thing.
It doesn't have to be perfect, but atleast you have comments you go through. Is it worth 5% of your time to do that?And. Push it forward. And, and also maybe it gets ignored. Maybe no one elsepicks it up on Twitter or anywhere else, but maybe there's a 5% chance it doesand maybe it inspires other people in other sectors like the exchanges or the,like foundations, whatever it, what other areas are pushing for regulationbecause I, I just hear this message a lot but I think it's, it's obviously notworth 50% of your time, but maybe it's worth 5% of your time to get out an mvpjust.
Offer the first, offer the olive branch,like be the first one to extend some, some sign of work. Because it is, it iswork and I don't think you would be happy. Like, in any legal negotiation,like, I wanna buy your company. Whoever presents the first draft of thedocument kind of has the advantage and that the other team is gonna beredlining the original doc presented by the, the, the bigger team.
So that's how I sort of think of it islike, present the draft, let them draft it up, edit as they want, but at leastyou have the framework out there in the world.
Michael: I don't know, I thinkthat's actually, I've never thought about that. So that's actually not a bad.And it might be something worth exploring, I think.
I'll say that like for me it's a biteasy cause I'm out there in public, I talk to people about what we're doing. Alot of people don't want anything, don't wanna be seen. They wanna keep theirsecret secret. They don't wanna become a target, they don't want to run intotrouble. So I think that's, that's part of what's disincentivizing other peoplefrom getting out there.
But I think it's certain something Ishould spend some time exploring. Cause it's actually not a bad. I'm skepticalit would move the needle given the current climate, but as you said, if it'sonly 5% of the time, it might be worth a shot. And if it's a lot of crypto istaking asymmetric bets and maybe it's another asymmetric bet that will pay off.
Mike: Yeah, and one thing I,I've certainly noticed is that I see a. A strong change in the attention ofpoliticians, right? You've seen this too, but just pointing it out is there'sless value and trust placed on powerful centralized publications and moreemphasis placed on podcasts and individual produ producers on Twitter,primarily.
Twitter. I mean, maybe you could argueYouTube as well, but it's like, Really, it's like, okay, if you,hypothetically, you put this thing out there, it's whatever, 10 pages of, ofGoogle Doc you, powered by AI for the, like you could, you could get it outthere, I'm sure pretty efficiently. And then you have a few other people whoare willing to put their name on it, given that they just.
At the bottom said this was reviewed andstamps with approval from these people. So all of a sudden you have kind of asnowball effect of credentials and then you just throw it out there, like tweetat all the people. Maybe it gets picked up. Maybe you do a couple podcasts specificallytrying to talk about this.
Number one, it gives you something totalk about which podcasters love. Like if you came in with emissions, like Iwanna talk about this piece of policy that our team wrote with othercoordination with other high frequency traders. And then it's like, I thinkpeople are dying for. Meaning and omission, like someone to come out and say,this is what we're trying to accomplish in the world.
We're trying to bridge the gap betweenpolicy makers and crypto because everyone is aware of this tension. Everyonedoes not want a war between BTC and U S D, but yet that's where we seem to be.So the more bridges that happened, the less chance there's just collateraldamage everywhere.
Michael: That's interesting.
And it's. I think another component ofit's, it's a really hard problem, admittedly. Mm-hmm. I used to, admittedly Iwasn't a great lawyer, that's why I did something else now, but I'll be thefirst to admit it. So, if you sat me down with a bunch of se circulations andsaid, wade through them and make them kind of make sense in crypto, it would behard.
I would definitely need the lawyershelps with that. But I think it's something that could potentially work. And Ithink you're right, the right avenue now has changed. It's not aboutcentralized mass media kind of getting the message out to people. I justwouldn't put it on TikTok. I dunno if they'd be crazy about
Yeah. TikTok probably doesn't seem likethe right platform now. Yeah, no, I think Twitter, Twitter's probably yourbetter bet. Cool. Well, I, fully support that. If I can connect you with otherpeople who, I've talked to, I'm happy to do that. We've covered a lot in thisconversation. We talked about high frequency trading, we talked about theinternational influence of China, Russia, the, the macroeconomics there.
Is there any area that we didn't touchon that you think is important worth giving some time to?
Michael: I think, I think onething.
We started as a, we're called DexterityCapital, and part of that reason is when we started trading in 2017 we tradedon Dexus and these things were broken.
They were just in really bad shape cuzthey were new. You would see things where someone was offering to buy a Coinfor a better price than the lowest ask. So a crossing order book. I mean,that's actually how we started. We could see that like someone was bidding 10cents for this thing and the another person in the book was offering to sell itfor nine.
Well, we just took both orders and madea penny. Right? And that was great. And we did that all day long. So Dex isdecentralized exchanges, dexterity. That's kind of how we came about. And weended up focusing on centralized finance and high frequency, because when 2018came along, DeFi cooled, and then there were kind of, that's where the moneywas, that's where the opportunity was.
And activity was one potential outcome.And the next problem for regulars tackle is how do you tackle DeFi? And that'sgonna be interesting. It has this double-edged sword of real transparency.Things are on chain, you can see what's happening. Nothing's really thatanonymous. Everything can kind of be traced back to some people.
We've seen as many of the recent hacksthat often they give back the money because they really can't dodge it. It'sjust gonna happen. Right? But it does kind of ask fundamental, interestingquestions. If you think of a, a Dow, a decentralized, autonomous organiz, Howdo you regulate a Dow? How do you sue a Dow?
It's, regulate or operated by thousandsof people who each have evoked, kind of distributed across the world. And thisis a problem that, I mean, I love the problem cause it's really hard andinteresting, but I have no idea how to address it. And I'm really curious howthe world's gonna see that when centralized exchanges are kinda gettinghammered.
Crypto's gonna have to look at DeFi moreaggressively. Hmm.
how does it work today? In, I, I don't know ifyou know the answer to this, but the first thing that comes to mind as ananalogy to DEX ownership would be real estate ownership. You could have a, afive story retail building owned by 150 people, none of which have more than 1%ownership.
Is there a parallel there? Like is therea board of directors or some sort of governing body? Most crypto projects havea foundation that's, Maybe three to five people who manage the, for the fundsfor a, a project like the Ethereum Foundation, Solana, et cetera, is, do youthink that sort of ownership structure or governing body structure makes sense?
Michael: It's, it's hardbecause these things are all direct democracies ultimately. Mm-hmm. Everydecision needs to be voted on by all the token holders. Otherwise, you havecentral. So when you were talking about a, a condo or buildings or what'sfrequently called a real world asset, rwa RWAs are cool, but at the end of theday, someone's gotta maintain that building.
Well, you can hire vendors, but then youhave to have a direct democratic vote where every token holder votes on that.How frequently are they gonna vote on this? Right. Are they gonna spend theirtime worrying about this? How are you gonna get your plurality or your quorumor whatever you need to get things.
So you end up again with RWAs, you endup with centralized management being pretty much necessary to make sure thatbuilding's plumbing gets sorted and people are paying their rent or their HOAsor whatever. And that's, that's still very tricky. So we're still in thatinfancy and I don't know what the right answer is.
And unfortunately we also don't knowfrom regulation perspective what the right answer is. Mm-hmm. And that's gonnadrive exactly how it operate. So it's really kind of a open space in terms ofwhat will happen. Very hard to take a bet.
Mike: Yeah. Yeah. Thechallenge with direct democracies, whether governing bodies or in digital land,is that it puts all the computational responsibility for every decision at theedge of the network.
Everyone has to do the full load tocontribute a vote that's valuable. And in lieu of that, you. You can all votefor people who do the work, and so therefore you can be part of five different,10 different DAOs or Dees, and you, your job is to vote for the smart people.And then those people spend all day, all night thinking about the best policiesfor that Dex, that's kind of the, I I think that was the, the innovation of theUS Constitution.
One of the, the major ones was likedirect, direct bureaucracy is just, Taxing, no pun intended, on every person.That's a part of it. If it's gonna work, it's like Right. That's the tradeoff.
Michael: Yeah. Yeah. I think,yeah. What we run into again, and, and this is US centric, but is again, thatHowie test.
Cause if you're holding that token, I'msure you've nominated some guy to go make the smart decisions. That sounds likecentralized management, and that means your token is likely a. And because weha are still working with this dated regulatory framework of what a securityis, it makes life hard. So I think you're right, like representative makes alot more sense for this application, but without clarity on how arepresentative Dao will be treated, it's a tough bet to make.
Mike: Yeah. Yeah, we neednude laws effectively. Oh yeah. Well, back to the last point this was this wassuper fun. I love the name Dexterity Capital. Sounds like you guys are workingon some really exciting stuff. And thanks so much for hopping on. This is areally awesome and dynamic inspiring conversation.
Are you writing publicly, personally, onTwitter or blogging elsewhere that you wanna throw?
Michael: No, not really. I wishI was, but it's just so hard to make time and there's, there's so many smarterpeople than you can write better stuff. Mm-hmm. But I should spend more time onit, for sure. Mm-hmm.
Mike: Cool. Well, we'll haveall the links for the company and the show knows, and thanks again.
Would love to have you back onsomeday.
Michael: Anytime. Thanks somuch.
Mike: All right, brother.Talk soon.