Bryn Solomon is the co-founder and CEO of MGNR, a quantitative cryptocurrency asset manager with arms in market making, discretionary trading, DeFi yield farming, and venture capital.

Our conversation touches on all these arms, exploring their key differences from traditional markets, sources of edge and opportunity, and the risks unique to the cryptocurrency markets.

A recurring theme within the conversation is how the pace of innovation in cryptocurrency presents both opportunities and risks. For bearing the burden of exchange outages, contract hacks, flash loan attacks, and outright scams, savvy traders can find opportunities in mis-priced derivative contracts, asymmetric information flow, and the occasional fat-finger error in crypto punks.

I hope you enjoy my conversation with Bryn Solomon.

Transcript

Corey Hoffstein  00:00

Okay 321 Let’s go 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  00:22

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 newfound research. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of new found research may maintain positions and securities discussed in this podcast for more information is it think newfound.com.

Corey Hoffstein  00:53

This season is sponsored by simplify ETFs simplify seeks to help you modernize your portfolio with its innovative set of options based strategies. Full disclosure prior to simplify sponsoring the season, we had incorporated some of simplifies ETFs into our ETF model mandates here at New Found if you’re interested in reading a brief case study about why and how. Visit simplified.us/flirting with models and stick around after the episode for an ongoing conversation about markets and convexity with the convexity Maven himself simplifies own Harley Bassman. Britain Solomon is the co founder and CEO of MGN are a quantitative cryptocurrency asset manager with arms and market making discretionary trading defy yield farming and venture capital. Our conversation touches on all of these arms, exploring their key differences from traditional markets, sources of edge and opportunity and the risks unique to the cryptocurrency markets. A recurring theme within the conversation is how the pace of innovation and cryptocurrency presents both opportunities and risks are bearing the burden of exchange outages, contract hacks, Flash loan attacks and outright scams. Savvy traders can find opportunities and mispriced derivatives, asymmetric information flow and the occasional fat finger error in crypto punks. I hope you enjoy my conversation with Brian Solomon. Bring Welcome to the show. Very excited to have you here. Thank you for joining me.

Bryn Solomon  02:30

Thanks for having me, Cory, nice to chat.

Corey Hoffstein  02:33

I suspect a lot of my listeners who haven’t gone as deep down the crypto rabbit hole as I have lately, probably unfortunately haven’t heard of you. So I would love to start with a bit of your background and how you got interested and involved and ultimately ended up joining the whole crypto world.

Bryn Solomon  02:53

Sure. I’m from a traditional finance background. And specifically I was working as a quant derivatives market maker for a number of years. I worked for four years in Hong Kong at a spinoff from optiver and then worked at another optiver spin off in Chicago for a few years. So I have a background in option market making Asian indexes like Hang Seng KOSPI, Taiwanese index Australian index, single stocks. In the US I was the head of a desk doing a lot of US Treasuries and fixed income products.

Corey Hoffstein  03:30

And before we dive into the crypto, I actually found an article you wrote a little while ago talking about your experience gambling and what I would consider to be a pretty professional manner, you would actually found a pretty interesting edge. I was hoping you could talk about that, because he spent six years doing it. It’s a pretty healthy chunk of time. What was the game that you found it and where was your edge?

Bryn Solomon  03:51

Yeah, that’s actually a fun topic. Throughout my history. I’ve found lots of little gambling edges like that among various lotteries and scratch tickets, and it’s surprisingly beatable a number of these things. And still today, I could tell you, some of us scratch lottery tickets that are beautiful, with varying degrees of effort and pay off. But this one in particular was a lottery in Australia called soccer pools or football pools if you’re European. And essentially what would happen is each week, if the big jackpot wasn’t one, it would snowball into next week’s jackpot, which is a pretty normal lottery mechanics. So one of the elements is you’re waiting for a very large jackpot so that the payoff is sufficiently large to overcome your expected loss on tickets. And then secondly, the interesting work about this was rather than like a Powerball where you randomly draw balls from a bucket and there’s no kind of a uniform distribution in the soccer pools. The numbers of the balls that are winning each week is actually determined by the result of soccer matches or football matches in Europe. So like in UK Premier League and Italian league and the Spanish league. And so I guess there’s super interest thing or fun realization I had is these betting markets on soccer are so professional, you can get really good odds from a lot of UK bookmakers on very specific not only to kind of match results, but score match results like, specifically will this be a draw? Will this team win by five goals or whatever it is. And so each week, I would scrape all of the data for those football matches and throw it into a big kind of brute force simulator. And I would simulate all the matches being played 10s of 1000s of times, and derive which matches were most likely to be drawn as the lottery winners that week. They’re just really interesting. Like, it seems obvious after you say it, but no one else was doing it. And I could tell from back testing the system that on the weeks where there would have been a pretty clear winner. Nobody had been implementing that system. So I was doing that for a while.

Corey Hoffstein  05:49

How would you figure out that your edge was eroding? What made you stop

Bryn Solomon  05:53

actually was not eroding. What happened was that particular lottery game got closed down by the lottery company. And nothing to do with me, right? Like the lottery still makes the same commission and week in week out. So my profit is just coming from other ticket buyers who are not using my system. But it was a dying in popularity, like soccer is not that popular in Australia, I think it was a bit of a sort of post World War Two game, although they were just revamping everything. So unfortunately, dead.

Corey Hoffstein  06:20

I love the article. Because as I read it, I was thinking to myself, Okay, this seems like a solvable problem. But man, I don’t have the data at my fingertips to go predict all these soccer games, and how’s he going to do this? And then when you turn to the betting markets, it was the most obvious and ingenious solution because you’re taking from the wisdom of crowds. So I thought it was a really fascinating article, I’ll make sure I link to it in the show notes. Everyone should give it a read. So all right back on topic back to market making back to the crypto world. So you did start ultimately, you transitioned from being a market maker in the US to you started market making cryptocurrencies. And I am curious, given your background market making in Hong Kong and market making in the US while you were ultimately doing it on different instruments, which do you think ultimately better prepared you for market making in cryptocurrencies? And why?

Bryn Solomon  07:17

Yeah, I’ll segue like how I got into crypto from the TRad fi markets was actually give credit to my co founder, who is our CTO at mdnr. And he is an ex colleague of mine from Hong Kong, was passing through Chicago, and crashed on my couch for a couple of weeks while he was doing an internship. And at that time, we just he started introducing me to Bitcoin. And then him and another friend had already started making some sort of rudimentary math and making buttons. So that was my first hook. Like, they pulled it up on a laptop and showed me this thing and get immediately all of your livestock going off. Like, let’s change that. Let’s change that. Let’s try this. So that was the evolution that yeah, so regarding Hong Kong versus us. So Hong Kong markets are marginally lower volume and less efficient than US markets. And so the place I was working out there, and the nature of some of that market making was a little higher touch, and also a little more, I would say, like, you need a little more hustle, you kind of need to dig around and sometimes find out what’s going on because things might not be as well documented by exchanges that the procedures are might not be as robust. And so I think that piece is really relevant to crypto, like almost the emerging market essence of Asian markets at least 10 years ago. And from the US side, what I learned at the firm there was that was further down the automation and technical spectrum than my first firm. So probably took a lot of elements of that, in terms of like really automating everything and systematizing everything and trying to remove the human from the equation as much as possible.

Corey Hoffstein  08:47

I find the concept of starting a market maker to be just an incredibly daunting task. I’m thinking about things like capital needs, and exchange feeds and hardware and redundancy management and regulatory oversight. And that’s, I mean, before we’re even talking about, do you have a model that works? So I’m curious, what did the sort of day 1021 process look like for you? What were those early days in practice?

Bryn Solomon  09:19

Right. So I think it’s interesting, a lot of people that I see coming from Chad, fie into crypto do come with a certain set of assumptions, and one of them is that there are big hurdles or big barriers to entry. Of course, if you were to try and start a market maker in the US, I mean, what would you need like $3 million of capital and a number of meetings with lawyers and setting up various corporations and licenses and it’s just like, to me that’s so ugly, that part of the world. And so, you might find it like really wonderfully refreshing to understand that like, crypto has a really low barrier to entry. Like, if you’re a quant person, if you’re like a college student and you think hey, I’d like to try trading, just build a model and just deploy it. So out there Xero example was back in the day on GitHub, there was an open source bitmex market making bot that was actually built by, I think, the CTO of bit Mex at the time, Sam Reid. And I believe that’s still on GitHub and still able to be forked. So as a day zero, like, literally, you could fork someone else’s market making but plug in your own API key, and deploy it immediately. And just start playing around with the parameters and coming up with ideas.

Corey Hoffstein  10:27

How much of it for you was automatically deploying in a fully systematized manner versus watching very closely hands on tweaking the models in those early days,

Bryn Solomon  10:39

I think, probably earlier on we had more human input, and it was a lot more about moving based on intuition. So that’s one of the big pieces of discussion at a quant trading firm. There’s this kind of trade off between just time to market and how thorough Have you been with your assumptions. And so myself, my personal bias is to lean a little bit more on the side of like, let’s just test him prod effectively. And so while we would like to go back and do hours of back testing and data analysis, like sometimes, if the hurdle is too high, it’s just easier to run a model with a very small amount of capital, live online and just have a look at what’s happening. babysit it, like just get a feel for what’s happening.

Corey Hoffstein  11:23

How does the structure of cryptocurrency markets versus say traditional markets affect how you approach them as a market maker, so not to lead the witness too much here, but I’m thinking about things like the fragmented nature of these exchanges, exchange risk, you might have the lack of regulation. So spoofing orders may be rampant, the different nature of market participants, perhaps more retail, less institutional, even things as nuances, different matching engines and order book algorithms on these different exchanges and contract specifications. Talk me through how all of that sort of factors into your approach as a market maker?

Bryn Solomon  12:04

Yeah, I think you’ve kind of nailed it hit the nail on the head on a few of the areas that so crypto markets are very fragmented, as in there are some number of major exchanges that do a reasonable chunk of the volume each. And so right now, we trade across about 20, centralized venues and a range of decentralized venues. So this is probably a reflection of of one regime of the kind of equity markets in the US, you know, pre reg NMS. I think when there was a lot of exchange consolidation, they did used to be maybe 15 US equities venues, and there was a period of time where various latency arms were ramping. So what else is different about crypto, it’s very retail heavy, like I don’t know the exact numbers, but I would liken it a bit more to the sort of Chinese equities market, which has a high retail participation. And so on the one hand, you have a lot of uninformed counterparties, which is kind of interesting. And on the other hand, because it’s so unregulated, there’s a lot of like, I don’t want to really use the word insider trading. But most market moving information is disseminated privately before it’s disseminated publicly. I don’t think there’s been any kind of legal ruling on that. And because it’s all secondary markets, I don’t think the law even necessarily applies. But by and large, a lot of the tokens you’re trading like they do represent teams who are building things, and various venture capitalists and friends of team will know things well before you know things. So that’s one thing to be aware of.

Corey Hoffstein  13:29

One of the things you often hear about its whispers, and I’ve never really seen it seem truly confirmed is that there are large whales out there who will from time to time, I guess, attack certain markets, or do big sell orders to set off these liquidation cascades or pump certain markets. How does that factor into how you have to think about market making?

Bryn Solomon  13:54

Yeah, sure. So in our models, we have a number of different inputs that try to take into account various metrics that may be more audible based metrics, or sometimes we might also incorporate higher level metrics, like open interest. So you know, on all the venues, you can sort of understand at least what’s the net leverage that people are carrying, this is more like moving into the realm of maybe even mid frequency machine learning type stuff. But I think if you believe that there are people intentionally manipulating these sorts of things, one of the most common trades or patterns would be that you would amass a huge position on a derivatives contract and then you would bang the spot contracts on which the derivative is based. And so that actually should be somewhat clear in terms of maybe a spike in open interest on the drums contract and then a sudden drop after the exploit has occurred. So there are some elements of that. I also think various firms and less so us because we’re a lot more automated, but various firms have contact with OTC desks and are always kind of pulse checking. What are your corporate clients doing? And they’re always buying so the only question is How much buying is happening right now. But that’s like a big piece of nonpublic information that’s probably quite useful as well. One of the big

Corey Hoffstein  15:07

features of traditional finance market making is latency minimization. So 15 years ago, everything was all about CO locating, and going down to the hardware. And FPGAs. I’ve heard that that’s just not necessarily the case in crypto markets because of the way they’re structured. Can you talk to me a little bit about how you’ve had to focus on latency? Or is it just really not a factor that you consider as much.

Bryn Solomon  15:37

So we do consider ourselves high frequency in the crypto space, and I believe we’re one of the faster firms that’s out there, you know, we have a pretty good hit rate on things we swing at. But I do think it’s a bit of a trap to come from a trade fi background and make the assumption that I like the number of guys I’ve spoken to that have mentioned the word FPGAs. Like in our first conversation, it’s like, totally the wrong way to think about it. So that timeframe is Nanos, or mics, and the crypto timeframe is much more in the millisecond realm. And why is that the case? Well, there’s a few factors, but one is that the venues are co located in different AWS centers all over the world. And so when you pass data between different AWS venues, most of the time, you’ll be using the internet or some comparable infrastructure and talking about, say, 100 milliseconds travel time between Asia and Europe. And so there’s just a natural jitter on that, which like the magnitude of that jitter, often trumps the difference between using C++ or some like mildly slower language. And then even worse, and even more important than the network latency, I think, is that the exchange matching engines are very inconsistent and very, under SPECT for the tasks they’re trying to accomplish. So every time it gets busy and trading, that’s when all the edge is happening, it’s going to be quite unreliable, in what order certain packets arrive with varying degrees of lag after the true event occurred. So there’s a whole interesting space there about how do you build your system to try to estimate what’s really happening when you can be quite certain what you’re seeing is old information. I think that’s a fun challenge.

Corey Hoffstein  17:13

One of the things that we see quite frequently is that when markets get chaotic, a lot of these exchanges end up having system issues. How do you deal with that, as a market maker who’s potentially carrying inventory risk? Yeah,

Bryn Solomon  17:29

there’s been at least two times in our history when we’ve had reasonably large blow ups, both times from something like what you’re mentioning, and so naturally as you if you’re trying to engineer a distributed system, you need to spend a lot of energy, pre empting. What could go wrong? I think everyone has spent a bit of time in high frequency automated trading understands, like that’s your biggest existential risk. Doing a few bad trades is like irrelevant. But the one thing that could blow you up in a second is some huge technical error. And so yeah, I wouldn’t say we historically have dealt with it optimally, because it has been us. But we just continually tried to iterate and anticipate different things that could happen. And specifically like to give you an example of what you are mentioning there, at least two times, on some of the busiest days, a number of the major exchanges have just switched off their matching engine, like literally Bitcoin, there’s too many liquidations, the books not thick enough to handle it, market makers have all cleared out, we’re just going to turn off the machines for a while and hope that everybody relaxes. So it’s how do you ever handle that? It’s quite interesting.

Corey Hoffstein  18:32

One of the constant themes that’s been persistent in all my conversations with people who are trading cryptocurrencies, from a quantitative perspective, is how much markets have evolved over the last several years. So hoping maybe you could touch a bit on that some of the evolution that you’ve seen, and how has that impacted the way you’ve approached models building, and maybe the half life of the models that you use?

Bryn Solomon  18:59

Definitely, we’ve seen this, I guess, I call it like a constant hamster wheel of which is pretty endemic to high frequency trading in any domain. In traditional markets, it’s the same thing. So you can never just actually this might be interesting for some level of retail audience, but there’s no such thing as just building a good model, and then just letting it run and making a ton of money. Like, every day, you’re not trying to innovate on that model, you’re losing ground to a competitor who is trying to innovate. And because it’s all very game, theoretical, any innovation you do make will be picked up by someone else’s data analysis very, very quickly. And if it’s anything that they can reverse engineer like the next day, they’ll deploy it as well. So it’s pretty hard to in some dimensions to get a unique edge. But yeah, I think we just like any good firm, we’re just very aware of the paradigm that like that’s what happens in low latency trading. And always just trying to innovate and iterate on our features. And yeah, we’d like it’s so easy to look back on, say if we had the models we have today, one year ago. We would like just destroyed it, we wouldn’t be crashing. And we would have said the same thing a year ago, about two years ago, or whatever it is. So it’s funny to see how fast things can change.

Corey Hoffstein  20:10

I was hoping you might be able to provide us with a concrete example of how that evolution has led to maybe some of the models you used to use no longer being production worthy. And maybe you could explain what one of those models was and why the edge disappeared?

Bryn Solomon  20:27

Yeah, sure. It’s easier for me to answer that question maybe more on like a discretionary prop trade basis, rather than speaking about our models directly. But like, I’ll give you an example of, we’re always sort of looking around for little side hustles and taking edge wherever we can get it. So one night example was in maybe January, when NF Ts were just starting to get really big. I think we were one of the earlier quant firms that understood that trend and spend some time having a look at the market. And even though volumes weren’t super high compared to normal financial instruments, we deployed some HFT market making algorithms on NF Ts and the bread and butter model. There was just these things are meant to be non fungible, but a lot of them are released in a series and they’re somewhat fungible. And the big example everyone might know from crypto is crypto punks, which have some sort of flow of value, which is like maybe $30,000 right now. So nobody wants to sell a pump for less than $30,000. But very regularly, maybe once a week, less than once a week. But once a month, somebody will sell a crypto punk with a fat finger. So there’ll be, you know, an off by one error, where they’ll be trying to sell for 30,000. And they’ll sell for 3000. And just the nature of the way that defy works, like when that offer is out in the market. If you’re the only person running a bot, like you’ll get to buy it first. So we used to do some things like that, which have just become a bit too high maintenance for us. But if I was like a on your own stay at home guy like, you can make money there for sure. And then another example, which is probably doesn’t exist anymore. But FTX the exchange has a series of they’re called Move contracts, which are effectively like a floating strike straddle on certain index. And at some point the strike gets ossified. But anyway, when they launched the I think they were just moving really fast and building a lot of things. And so they didn’t have time to micro optimize all the different markets, they were market making. And I’m pretty sure the market maker there was running a flat vol surface. So if anyone who has traded volatility before knows, there’s a reason that there’s such a thing as a vault smile. And there’s a reason that extreme strikes traded higher implied volatility at the money strikes. And so you should just arbitrage that market maker all day on the move against debit, where there was a more liquid and more kind of correctly price ball surface, but it was very low volume stuff. So wasn’t a big win.

Corey Hoffstein  22:51

One of the more recent evolutions we’ve seen in cryptocurrency markets is the rise of defy decentralized finance. And with that have come all these decentralized exchanges that are being powered by automated market making smart contracts. How does the opportunity set for someone like you change if the entire crypto world moves away from centralized exchanges to these decentralized ones?

Bryn Solomon  23:21

At a high level, I think sci fi is sometimes borrowing elements of trade fi and in the same way, sometimes as a trader, your pre existing knowledge set is applicable here. And then concurrently, there is I believe, some new primitives being created that have been enabled by cryptography and by programmable money. And so I think the way that the opportunity set evolves for us here is you probably need a much different hybrid of people with markets knowledge and people with engineering and building knowledge. And at least Yeah, I found for us, that’s been one of our big edges in now defi side, that, you know, we’re a team of about 10. Now, and maybe eight and a half of us are engineers. And so everybody can pick up a smart contract and have a look at what’s being built. Every token you trade represents something being built, and why is it being built? What does the architecture look like? Read the code? How can you like, have they missed anything? Like, are there any weaknesses in the way the code has been structured? And so that’s one element. Yeah, maybe a more simple thing to illustrate, too, is just in general, if you are trying to trade on defi protocol, like uniswap, as opposed to a centralized exchange Aetherium blocks are only stamped every 13 ish seconds. And so in a way that sort of speed bumps a lot of orders, almost like the iEX like Brad Katsuyama kind of idea. And so, it introduces a whole new range of interesting game theory, that we’re not always competing on price time anymore, we’re often guaranteed to be in at the same time and so, in what other ways can we be tricky to try and compete for the right opportunities?

Corey Hoffstein  24:59

I also wonder, in the case of decentralized exchanges as they are now, you lose a certain amount of information because the order book no longer exists. You can’t post certain types of orders at different parts of the order book. Do you think if those centralized exchanges were to go away, and I know I’m talking about something very extreme, but the entire world moves to decentralized exchanges, there is still sufficient opportunity to warrant running and market making firm.

Bryn Solomon  25:29

Yeah, so I actually think that the kind of final state of AFM is to just end up back at the same place as Knott Central, but like limit order books as we understand them today. So this whole amm concept, I feel like it’s novel. And maybe there are some fun elements or lessons we can take from this. But my personal view is that in five years, we might look back on this period as an anachronism and say, have a romantic rosy glasses. Oh, that was fun. But boy, am I glad we’re all back on normal limit order books now because there’s a lot more information and kind of you’re able to do more things as a participant,

Corey Hoffstein  26:03

maybe just really quickly for listeners who aren’t aware of what amm is, or can you give a quick high level definition of what an amm is?

Bryn Solomon  26:11

Right. So MLMs are an idea that’s meant to attempt to replace a central limit order book, where all of the liquidity providers, which can be anyone can choose to just put their money into a shared pool. And so there’s this big pool of funds, and then all trades happening in a certain instrument can occur according to some mathematical function. So before you execute a trade against this pool, the price is known. And you can choose to execute the trade against the pool or not execute. But in this way, we kind of remove maybe two layers of rent seekers arguably. So exchanges are no longer relevant or necessary in this paradigm, because it’s always party to pool, you would call it like someone doing a trade against the pool of individuals who receive the other side of that, in addition, the protocols can kind of charge the fees that an exchange might have used to have charged so the liquidity providers in this paradigm can almost kind of receive the profit of like a CMA by interacting directly with other people. And then maybe the other rent seeker you remove is guys like us, which is like a high frequency market maker who tries to get in there, between various order flows,

Corey Hoffstein  27:23

to your point about decentralized exchanges converging to centralized exchanges, I think in the most recent version of uniswap, that was proposed, what they’re actually saying is in this liquidity pool, you can choose where you want to provide liquidity and how much liquidity you want in different places within the pool, which I think is interesting. But it highlights, again, just how fast this space moves, and how quickly things can change within these protocols. I’m curious how that inhibits your ability to truly automate everything, how much has to ultimately end up being somewhat discretionary because of the pace at which these cutting edges are changing. And while I would expect there to be a lot of opportunity in the cutting edges, it also has to invite a tremendous amount of friction.

Bryn Solomon  28:13

That’s a very relevant question. And so the further you move into the defi world as a trading firm, so from our perspective, the more that we kind of have to deal with that particular set of trade offs when we’re deciding about resource allocation. And so how persistent do we think this paradigm will be. And even if a paradigm might be persistent, sometimes it’s quite obvious that the edge is going to dissipate, or like disappear very quickly, because it’s just too clean and too obvious. And there’s no identifiable like mode to our deployment. So yeah, that like as a guy who has a mixed background of automated and discretionary trading, like I love the element of defy of being able to, and it being very necessary to keep making discretionary calls and set up sort of little scripts that you know, will work for three days, and then won’t be relevant anymore. And it’s just like a lot of very fast moving iterations.

Corey Hoffstein  29:04

So I want to actually talk about some of those discretionary ideas, because I know as a firm, market making still remains very core to what you do. But you told me you also have started introducing some discretionary trading, some defy yield, farming, and even some venture investments. So I want to touch on all of those, starting with the discretionary management. So I would love to know, what sort of opportunities are you ultimately looking to identify here when you talk about putting money to work in a discretionary capacity?

Bryn Solomon  29:36

Yeah. So start with the idea about just maybe like, as a very, let’s call it like as a power user of defi and crypto in general. We have probably like a very nice network of information, and were very aware of things that haven’t been built yet are being built are being stepped up. We help teams actually kind of spec up a lot of the architecture or concepts for things that they’re going to build And so naturally, I suppose what you start to see is, it’s quite easy to skate to where the puck is going. Or when I say quite easy, like, we have an advantage there, that’s probably an advantage over the average market participant. And so a lot of the discretionary trading operates in that realm. It’s just kind of like a very easy trading crypto that has probably been working for two years is when protocol is about to announce news in two months. And you know, it’s going to be big news, it just doesn’t get priced in two months in advance, people just wait till the day of the announcement, and then there’s just a 10% pop. And so this is all just very, like penny stock II kind of behavior. And I don’t think that will be a persistent market behavior forever. But it certainly is the case right now in crypto. And it’s quite easy to get those bets, right, sometimes.

Corey Hoffstein  30:43

Can you walk me through the lifecycle of a discretionary trade? Where are some of these ideas coming from? How are you choosing to size the positions? And when are you taking the trades off?

Bryn Solomon  30:54

So regarding sizing, like that’s just very, not systematic at all, but we do, we ring fence a small portion of our portfolio, maybe 10%, of our AUM at the moment, and we play around with various discretionary ideas. So maybe a month ago, alpha hamara, which is a protocol and the token is alpha, they had been kind of publicly projecting that, hey, we’re about to release our new revamped v2 And you need to own these alpha tokens to have like a boosted yield multiplier on some of the activities. And because we’re just quite involved in the space, we know that people are going to want to use that product. And we know that people are going to want to get that yield multiplier. And so you can either wait until the release and then go and buy the token, which a lot of people are probably going to do or we can just buy the token today and wait for it to slowly kind of find fair value much higher. And so yeah, we do trades like that. And I find very often in some of our discretionary trades in terms of full lifecycle we use the event as the catalyst for exit liquidity so most of the time sell the news is the right idea that was the same with the uni swap v3 launch. I’m pretty sure it was like close to all time high on that day and then it’s just been Clift I’ve ever since

Corey Hoffstein  32:03

I want to transition to talking about defy yield farming. And maybe I’ll start with a couple open questions. What is defy yield farming? Why do these farms exist? And how is this all not just a big scam?

Bryn Solomon  32:18

I think the kind of knee jerk or first instinct reaction from a Trad fie mindset is that it is a big pyramid scheme. And unfortunately, we missed what seemed now the summer of 2020, or like defi, summer, when there was a lot of opportunity in yield farming. And it’s because I personally had to skeptical open outlook about it. And so I was aware of the of its existence, but didn’t do enough of a deep dive to pass over the edge. But in since about the start of 2021, we’ve been more involved in yield farming. And there’s a number of kinds of frameworks or parallels, you can propose to try and understand what’s happening. But the one that I like to use is, at the heart of it, companies are just rewarding users with equity in the company. So when I say users, it’s kind of participants in a protocol, or people who complete certain behaviors or demonstrate certain behaviors that the protocol wants to encourage. And from that perspective, it’s not necessarily just a strict like pyramid scheme or scam. So if we look at like the 2010s, Silicon Valley venture era, like what’s a normal venture companies lifecycle look like they’ll go and raise from venture capitalists, and they’ll raise maybe a lot of money. And then they’ll go and sort of burn the cash by giving away free things to the users to try and encourage growth in usage. And often, the firms are just grooming vanity metrics on various websites, maybe they just want to show number of daily active users monthly active users, even though it’s not always clear how that converts to revenue in the short term. There’s some sort of vague assumption that in the long term, if we just get the people will be able to do something with that mindshare. Yes. Similar paradigm in defy yield farming, except the protocols are kind of have full control over their own equity, their own token, and they use that token to directly drop it to users who are participating in the protocol.

Corey Hoffstein  34:09

It seems to me like when I look at these different projects, some of the greatest rewards exist with the most nascent projects, right? Again, these projects are trying to incentivize certain behavior, and they need those early adopters. And they need that liquidity created for their governance tokens to be able to be traded with other tokens out there. That also means that if you’re going for those greatest rewards, you’re going into an area of greater risk of failure, if not outright, higher likelihood of rug pulls and scams. Curious how you think about weighing this trade off between the opportunity for reward and the risk that goes along with it. Yeah,

Bryn Solomon  34:50

you’re correct that the effective yields on defi farms tend to decrease over time and for the reasons that you mentioned because there’s a protocols need to incentivize adaption early and then the idea is that they try to wean you off that sort of free money and and hope that there’s real usage of the protocol. So in a lot of ways, it is the case right now that the most active yield farmers and the most profitable, it’s a very high maintenance business. And actually, there was a report released by Nance and I think yesterday Nansen is a on chain analytics service. And I think about 40% of your farmers are out of the protocol within a week, or within 24 hours, or some really short timeframe. And so it’s certainly not the case that people that everybody is just putting lazy money to work and kind of accepting these decreasing yields. Like, if anything, that data shows that money is not loyal at all. And most of these people maybe don’t care about using the protocol, they’re just there to milk all of the early rewards, yields, and then they’ll be off to the next one. And so, yeah, that’s kind of the that’s the life of a very active yield farmer just hopping between opportunity to opportunity and making lists of upcoming opportunities. And if you can get in there from moment zero, it’s definitely the highest reward omissions. Some other ways that we think about rug risk, as you mentioned, like the earlier you move on something, the less Lindy it is, which is a, when something is linear, it means it’s existed for a long time. And if it’s existed for a long time, it probably has a lower probability of being hacked. Because if it could have been hacked, it would have been. So yeah, there are other heuristics that we use early on. And one of them is if the team is doxxed, which means their personalities are known, or they have a lot of online reputation, it’s normally a good sign. Because if you burn people on a project that’s linked to your online reputation, that’s going to be your last project in crypto. And so good engineers would like I’m not going to intentionally rogue people. And another good sign that we look for is just, this is a lazy method. But if a large amount of money goes in from somewhere else, we have a pretty good idea of most of the big wallets on chain. And so we know where the money is coming from. And so if Firm XYZ is willing to just throw $100 million into something on day one, they’re not doing it, because they think there’s a big chance they’re gonna get robbed, there’s a reason they’re very confident in the security of that particular protocol. And so you can just follow,

Corey Hoffstein  37:14

think the defy yield farming space, there’s a couple of different ways in which it can be played. One of the common ones is just sort of liquidity, mining, putting your money into these automated market making protocols, earning some yield, but also often getting rewarded in some sort of token for doing that. One of the risks that you take on when you are part of these liquidity pools is this idea of impermanent loss, I was hoping you might be able to explain what impermanent losses and how you think about accounting for it when you’re considering these different trades.

Bryn Solomon  37:49

Yeah, one way you could kind of understand what impermanent loss is, is just, if you imagine you’re a market maker on a foreign exchange pair, so JPY USD, and there’s a big one way move in the currency, which is otherwise normally somewhat stable, the market maker, if they don’t have a kind of very strong momentum factor in their model is just gonna get totally steamrolled, because they’ll keep selling or keep buying in one direction. And so nomenclature of impermanent loss is a bit unfortunate, because it’s very much not impermanent in the same way that maybe if you sold a straddle on an FX pair, and there was a large one way move, mark to market, you have a large loss. And you could sit there in your chair and say, oh, boy, I hope it comes back. Because if it comes back, I won’t have to realize this loss. But this is all just mental accounting, impermanent loss is actually real loss. And it’s real at the instant that it occurs. And people sort of hope that there’s some sort of cointegration in these pairs, and that everything comes back and they can get out unscathed.

Corey Hoffstein  38:47

As I’ve sort of evaluated this landscape, the thing that stays in the back of my mind is that, once again, it seems like diversification is just the most prudent form of risk management, we can diversify across the protocol. We’re using the different pools, we’re in the types of reward tokens that we’re collecting. That said, when I look at this space, and I talked to participants, one of the key features seems to be leveraged, that people are borrowing against their deposited capital, and then taking that borrow and depositing it somewhere else, and then often borrowing against that deposit, folding their capital 234 times over, which introduces potentially a huge amount of leverage into the defy system. And I would see that as a systemic risk. I’m curious as to how you think about the stability of the defy system as a whole with all that levered participation occurring?

Bryn Solomon  39:47

Yeah, nice concept you think about? You’re correct, that there are a number of ways to sort of re hypothecate your capital within the defi ecosystem. And yeah, people use the expression folding that you mentioned, which Yeah, you know, a 2x fold As much as b to x leverage. So there actually was a time when we cared a lot more about this than the overall stability of the whole ecosystem. And we were out there looking for our Michael Burry trade and trying to identify what’s one of the sort of underappreciated really like weakly collateralized protocols that will get totally screwed if there’s a big market sell off. And we’re recording this in sort of June. But on May 19, there was a massive market sell off that week, Bitcoin was down 50% from the highs and a lot of the defi tokens even more so. And I think one thing that was interesting to note is that there was no absolutely massive defy protocol pull ups. And if you can survive a reasonably quick 50 to 75% drawdown, it certainly indicates there’s maybe there won’t be a total systemic failure. Although definitely from time to time, there are various types of hacks, flash crashes, or Flash Crash events. The other thing I was gonna say about the way that a lot of these protocols handle liquidation is that it’s often open source and anyone is allowed to perform the liquidation task. And if you liquidate someone else who has negative collateral or has less margin than they need to have, you actually get paid a reward by the protocol. And so I think that’s quite neat as a concept, and I quite like it, if you look at the way that centralized exchanges work. Now, in the trade fi system, clearing houses are, I mean, I think they have a number of members, but the whole mechanics of it, the whole operation is very centralized. And sort of, we’re gonna liquidate you, and we’re in charge of this operation. And that’s probably closer to a single point of failure than the defy paradigm, which is, hey, we’ll open source this code to operate a liquidate a bot. And your bot can just always be looking, if somebody has less margin than they need to, your bot can just hit that request, and you get to collect the reward and then liquidate the person and take on that

Corey Hoffstein  41:52

position. When I’ve messed around in this defy space, one of the things I’ve noticed is that it is just incredibly expensive, particularly in the last six months. It seems like on the Ethereum blockchain gas fees have gone exponential, and they often are highest when the blockchain is most congested. When there’s all that activity going on. How do you think about accounting for these transaction fees in harvesting these defy opportunities, versus maybe the ability to be a market maker on a centralized exchange where you’re actually earning a rebate.

Bryn Solomon  42:33

As a matter of fact, right now, we’re in a little bit of a local Valley for Aetherium phase. And I think that’s been catalyzed by the recent sell off. So people are less euphoric, performing fewer operations on the chain, and therefore the title kind of gets false. I think with things like rebates and gas fees, it might be a bit of a trap, at least from a retail mindset to think of these things as like getting paid money or paying money. Because from a quantitative perspective, all we care about is when we perform an operation that we come out with more than we started with, and fees of any magnitude in any direction I just a function of that equation. And in fact, they’re a very simple linear function. So gas fees can be input into anything you’re running. And if they’re high, you trade less often. And if they’re low, you trade more often. And likewise, on a centralized exchange, rebates are just part of an algorithm and you can be tighter if you get rebated more, and you need to be wider if you don’t get rebated more,

Corey Hoffstein  43:30

most recently, you introduced a venture arm into your business, which is very different than core market making. What do you see as sort of the perceived opportunity set for you?

Bryn Solomon  43:43

Yeah, thanks for asking. So yeah, it’s been very interesting for us and very successful so far, which is, I would say, more a function of market beta than our skill, per se. But it was a very natural transition for us as we started to be more involved in yield farming defi trading, it’s necessary to stay abreast of what’s happening in the defi ecosystem as a whole, and what things people care about and what things people are trying to build. And maybe you even start to move further upstream to be in the right position for various yield farming opportunities or defy market making opportunities. And you just get to a point where you think, Well, we know these guys are building XYZ, and we know it’s going to be good. Let’s invest in this project with these guys. And I think probably a more unique skill set that we bring to crypto ventures than maybe the majority of other crypto bases is that we’re actually very kind of deep into building systems quant trading engineering. So I feel like we have very meaningful input in terms of actually looking at how things work and being able to think as an adversary, and how would we exploit this what sets of incentives are poorly designed? Like that’s kind of our bread and butter from the other side of the fence. So when we’re on this side of the fence, we can help design these things optimally.

Corey Hoffstein  44:56

How does venture in the crypto space differ? are from traditional venture, a number

Bryn Solomon  45:02

of things that are a little different. And so just want to preface this with the caveat that I am not from a venture background. But you know, I’m reasonably aware of how things work at a high level in the Silicon Valley world. So I think one notable feature in crypto is that deals happen really fast and really far upstream. And that’s because just a lot of things are still not very regulated. And regulation is a friction and it slows things down. And even you know, there’s anonymous avatar builders who raise money for projects, because you can see everyone’s money on a chain, and it’s all sort of traceable. And then so people are more comfortable with that idea. And, and so some deals happen without contracts, then some deals happen with very little information. And it’s more about like a wing and a prayer. But it’s not just hoping it’s sort of, hey, we understand these people, we know what they’re trying to build, or they have a pretty good reputation socially. And so anyway, it’s super fast. So there’s a set of venture firms in crypto that are the more traditional firms like Andreessen Horowitz level, and I think they’re actually probably excluded from the majority of opportunities, because like that burden is going to be too high. I mean, if you ever mentioned the word like diligence, or data room or something like that, like you’re probably rolling out 90% of the kind of fast moving venture opportunities. And the other thing that I think is underappreciated, and I haven’t seen a lot of writing about it, but I think it will become more acknowledged is that the time to exit is really compressed. And so probably in traditional venture, the majority, or a very large proportion of your risk is a function of time to exit, you’re betting on a team. And you’re hoping that in maybe three to seven years, you can get to some point where you get acquired or you get IPO, and you get your exit liquidity, because all of the crypto markets are already secondary markets, and they are super liquid, and mostly unregulated, or under regulated. When teams make these deals, they kind of just start vesting your equity or your tokens, sometimes three months after the deal is struck. And the more sustainable teams or projects that have big ambition, the vesting schedule can still be a long time, it can be a four year vesting schedule, but it’s kind of linear, so you can de risk along the way. And you have immediate market liquidity to sell into if you choose to de risk. And yeah, so I do think that I’m sometimes very skeptical, but I do think a lot of venture is just kind of fake it till you make it paradigm. And it’s a lot easier to fake something for three months, or one year than it is for seven years. To get to that IPO point.

Corey Hoffstein  47:34

I think one of the most interesting ideas about this space that you actually clued me into was that your exit can be in many ways programmatically implemented through a smart contract. So instead of waiting five to seven years for an exit, or then perhaps trying to find secondary market buyers, after a certain amount of time for your equity, you can actually have the case where you’re getting emitted these tokens. The second, the firm spins up, you can see programmatically on the blockchain, your actual emission schedule for earning this equity. I’m curious how you think that changes the risk profile. These are high risk ventures. But on the other hand, your time to exit is perhaps faster, you can evaluate how you’re going to be issued these tokens in a programmatic manner. You don’t have to necessarily trust the founders to find an exit for you. How does that change the risk dynamics of this type of venture?

Bryn Solomon  48:36

I say that definitely de risks it a bit. Not every team implements an automated kind of emissions contracts. But yeah, maybe even that’s a slight Alpha leak, like if a team is going to build some sort of smart contract that automatically vests you in some sort of linear fashion. It’s a good indicator that they’re stronger engineers than a team that would say, oh, no, every month, we’ll just like, send you the tokens manually. It’s a minor red flag, or kind of at least a minor indicator about engineering abilities. But yes, so if you have a smart contract, and you’ve invested via crypto out of a certain wallet, and then you can see that programmatically, that same wallet will receive the tokens that you purchased or you invested in, it definitely minimizes the founders ability to kind of just disappear with the funds. And so as a risk factor, that’s one thing that you can kind of take off the table, or at least he’s less likely.

Corey Hoffstein  49:27

What do you see as your core edge in the venture space

Bryn Solomon  49:30

has sort of touched on it earlier, but I think there are probably not too many very technical engineering background kind of trading firms that also really maybe understand defi venture and the way that the whole defi ecosystem works together. And so we’re a team of 10 and maybe eight of us are engineers by background and so we I feel like most of the time when we participate in venture deals. were substantially more thorough about details and probably accept a higher burden in terms of how things are actually going to work rather than we’ll just figure that out when we get there. Because we do understand what engineering challenges look like. And we do understand the scope of building various things. We’ve also been involved in building a number of on chain applications ourselves to facilitate trading and also working with other projects. And so almost an additional resource in terms of thinking adversarially as a trading firm, and thinking technically, as a group of engineers.

Corey Hoffstein  50:31

It’s pretty well known at this point that crypto markets go through incredibly violent swoons are we’re recording this in early June, and just a couple of weeks ago in May, we saw some very strong sell offs, though, historically, not even close to some of the strongest. Can you walk me through how the team navigates these types of periods? And I guess this will have to be a multiparty because I’m curious as to how it affects things like your market making versus the discretionary trading versus the defy yield farming.

Bryn Solomon  51:03

Right. So yes, specifically on that may 19 event. So the first comment I’ll make is, we’re very far down the automation end of the spectrum. So most of our bread and butter quant systems are, do not require human touch or human oversight whatsoever. And so definitely, when the market gets busier, we like to we have a number of dashboards and metrics that we’ve built to help us keep a good eye on what’s going on. Naturally, we want to just make sure that things are so out of sample that the behaviors are getting strange or that we’re accruing really large positions on odd instruments that we don’t expect. But yes, this may 19 sell off that there comes a point where maybe my discretionary instincts steps in a bit and and you combine that with a number of liquidations that are happening, which are you know, these are all feeds from the various exchange websites. So you can keep a track of at what rate liquidations are happening and what magnitude of liquidations have happened. And if you make a reasonable assumption that liquidations are undesirable trades. So for sellers, at some point after a massive amount of that happens, and then starts to slow down, it makes sense to step in discretionarily and pick up some data, even if it’s just for a short term view. So we did a bit of that and we have toggles and parameters in our system where we can default is that we’re 100% Beta hedged, but we can start to dial that down. And so just scale into some beta. And that’s in a very nice automated way, because our bots are still just going to do their normal thing. But we’re just tricking them into having a false perception of what position they want to occur. Besides that, I think one of my just normal reactions in a very busy market time like that is to just pull up Twitter, I still think a lot of good information comes out of Twitter, before we realize it despite the number of systems and tools that we’ve built. And so there are people who will just tweet, Hey, there’s this weird instrument on this exchange, and it’s just gone down 70% For no reason. And you can just go and sort of pick up little snaps like that sometimes, if you’re aware of the right info that’s out there.

Corey Hoffstein  53:01

Very same week, or shortly thereafter, there was a documented flash loan attack on, I think is the project name is pancake bunny, which is hard for me to say seriously, but such are the names in the defi space, and it was about $200 million plus was taken through this flash loan attack. And this is something that has been recurring a bit in the defy space, I wonder how you think about those sorts of different hack risks or flash loan type attacks, and how you try to protect yourself from them.

Bryn Solomon  53:35

Yeah, so that’s a very big risk for yield farming. Like that’s the primary means of a hack. Using a flash loan, people can amplify the amount of capital at their disposal immensely, almost infinitely. And the reason why this is possible is because the way that a chain like Aetherium works is that it’s a giant computer. And it’s doing all these different operations. And as long as the books balance at the end of every operation, it’s deemed a valid operation. So it’s just kind of doing some high level checks, that money didn’t go missing from the system or incorrectly transferred between wallet. And so someone came up with this primitive called a flush loan, where you could borrow an infinitely large amount of money, do a whole bunch of trades that are arbitrage is where you come up with more money than you started, then repay the loan. And you can do it all within the same set of operations such that the, the Aetherium chain recognizes that as a valid output, because you’ve paid everything back that you needed to and you got to keep all of the cream that you exploited. So yeah, I suppose one thing to just be aware of is that maybe when people are looking at protocols or designing systems, they need to understand that having a large amount of capital might not be a barrier to getting hacked. You could build a system and say, Oh, but someone would need $2 billion to do that. So it will never happen. Well, you can flash one $2 billion really easily, but somewhat easily. And so you need to kind of understand that that’s a very real risk factor.

Corey Hoffstein  54:58

Bring last Question for you. And this is a question I’m asking everyone who’s coming on this season, though, perhaps a little less relevant to you at the moment because I know you’re in a country that is, from a lockdown perspective, far more open. But the question is, generally, as the world seems to be coming out the other side of the COVID crisis, what are you most excited about? And perhaps we can tie this in? Because I know you just went through a 10 plus day quarantine, during quarantine, what were you most dreaming of doing it as you were waiting to get out the other side of the door.

Bryn Solomon  55:33

As you mentioned, I was just yeah, just in a mandatory hotel locked down for three weeks it was so 21 days, honestly, I had an intention to sort of do more regular daily exercise, so little YouTube workouts, but the reality is, I think I just did a lot more work and had a lot less fun than I probably should have. And started to feel like my body was my muscles were atrophying a bit. And so the thing that I was really looking forward to is out my window every day, there’s a very nice bike path along a river that I could see out of the hotel. And I just been sort of making this plan like the first day out, like I’m going for a long walk or a long bike along a path like that. Just moving my limbs. Sounds simple, but it was very satisfying.

Corey Hoffstein  56:14

I love it. Brian, thank you so much for joining me, this has been wonderful. Thanks, Cory.

Bryn Solomon  56:18

Appreciate it.

Corey Hoffstein  56:24

If you’re enjoying the season, please consider heading over to your favorite podcast platform and leaving us a rating or review and sharing us with friends or on social media. It helps new people find us and helps us grow. Finally, if you’d like to learn more about newfound research, our investment mandates mutual funds or associated ETFs. Please visit think newfound.com. And now welcome back to my ongoing conversation with Harley Bassman. There’s currently a lot of focus in the marketplace around structured return on equities. And I suspect that’s largely because most variants and investor portfolios comes from their equity position. But I’m curious as to your thoughts as to what other asset classes we might think about where convexity could provide a really useful profile for portfolio construction,

Harley Bassman  57:17

with rates having come down significantly from where they were a decade ago. I mean, we were in the 567 range for 20 years. And we’re at much lower rates now. And when that happens, people have desires, especially boomers, people in retirement, to have more income, they prefer to have their return given to them via a check every month or every three months, instead of having to go sell pieces of the stock or the asset off over time, Wall Street has found ways to do that, where they’re converting theoretical potential capital gains into more certain current income, and at a fair price for the underlying embedded options, and then a fair fee for the dealer. That is not a bad trade. I personally have bought many structured notes from Wall Street, they’ve made their profit, as have I, and we’re all happy. And this is because there are people out there who sometimes structurally need to buy insurance, they need to buy protection. And Wall Street fills that gap by taking the protection that one company might need and giving it to them via chopping it up into pieces. And having retail sell those pieces into a structured note that then bleed through to the underlying buyer, that is a public policy good of moving risk back and forth at a fair price. And thus, a lot of the structured notes that Wall Street has created are pretty good, there are some by the way that are not. And that becomes more challenging. And some of these things can be really tricky. And this is where a savvy financial adviser is actually worth his weight for the commission. Because when you start to put digital options, knockout options, binary options into these packages, they become difficult to value and to appreciate the risk, but in general, structured products can look pretty good. And the good thing now with Dodd Frank, is that in the prospectus they must disclose all the information they must disclose the fair value they must disclose the profit markup the Commission’s and that makes nothing is as good as disinfectant as bright sunshine separately. Let me add another concept over here which is basically adding in optionality convexity to other assets. So it could be credit, it could be gold, it could be rates, it could be we already have equities, interesting concept over here, because usually people are you was to do options on straight up equities. But it’s kind of unknown to trade options on bonds, or gold or credit. So offering that kind of asymmetric convex profile, where you could have much larger gains with control limited losses is a very clever idea. And our longer term goal is to broaden our menu to offer asymmetric returns convexity optionality to a wide range of underlying instruments. So basically, well, I don’t want to push too hard here, but you can buy a total menu of simplified products and complete your portfolio and never leave our house.