The FTX Podcast with Su Zhu Yield Farming Explained Risk and Rewards

Hey Tristan, thanks for having me.

Sure. So, based in Singapore, run a crypto hedge fund called Three Arrows Capital. We trade large caps, midcaps, also do a lot of primary investments these days as well, kind of from both short-term to long-term trading, and also a client of FTX. We trade pretty actively on the exchange.

Sure. I guess 2013, late 2012 or so, started hearing about Bitcoin and found it interesting. People were using it to move money around. I was in Hong Kong at the time, also in Macau, and kind of got interested in just the way that it could be used as cross-border movement. I think the arbitrage opportunities then were also starting to get interesting but not too scalable yet. You could buy it on Bitstamp, sell it on BTC China, do that kind of trade. At that time, concurrently, I had been starting Three Arrows, which was at that time mainly focused on emerging market currencies. So kind of ended up focusing more on that, and crypto started doing less of just because it was not very scalable. And then kind of picked it back up once we saw it redeveloping into a real market after the bear market ended. I think that, you know, thinking back to that time, it was quite a deregulated time compared to even now, right? Because you even had China spot exchanges where you could deposit fiat from a bank, you know, China Construction Bank straight into an exchange, whereas seven years later you still can't do that. So I think in some ways, the whole space has developed quite different to what I would have imagined back then, but definitely an exciting space.

Sure. That's a good question. No one's really asked me that. But I majored in math at Columbia. At that time, I wanted to go into academia. But I did an internship at an investment bank my sophomore year, and I didn't really know what I wanted to do that summer, but I had a friend in a class who said if you go intern at an investment bank, then they'll pay you $10,000 for the summer and they'll also let you travel a bit. You can kind of get to see some things. So that was compelling for me because I hadn't really traveled too much on my own before, and I thought the pay was decent. So I kind of just went into it blind. They had me half in Tokyo, half in Hong Kong at Credit Suisse. And at that point, my boss in Tokyo really liked me and he said, 'Why don't you join full-time? But I don't want to wait two years, so I want you to graduate early.' So I ended up taking him up on that. That was in late 2008. And then he had to lay off the whole team about four months after I joined because of the financial crisis. The whole book blew up. It was equity derivatives trading. But I was really fortunate then. So at that time, I was like, okay, do I go back to the states or go somewhere else and try to find a different path? But I managed to get a junior trader job at Flow Traders in Singapore in early 2009. They had just had a very good year in 2008 doing ETF arbitrage, and they were opening a Singapore office, so they just needed people to show up there. They had only four people at that time. So I think that's where I actually learned a lot of what I know now, which is market neutral trading, directional trading, special opportunities, special situations type trading. So I'm really fortunate that Flow gave me a chance to work for them. I did a couple years there, then moved to Hong Kong, worked at an investment bank prop trading group for about a year, and at that point just decided to start my own firm with my partner Kyle.

Right. That's right. Yeah. And he had worked entirely at banks doing similar type roles, but yeah.

It's quite different. I think in FX, our strategies were much more focused on the market structure. There was a multi-dealer market in FX and a multi-venue market, so you had OTC versus listed, and top-tier banks versus second-tier versus third-tier banks. One of the reasons we started Three Arrows in 2012 was because we realized that there were a lot of ways people were mispricing liquidity or miscalculating how liquidity should be redistributed. There were a few trends in our favor back then. One was that all the EM desks wanted to move from voice to screen, creating single bank platforms with a GUI. Next, second and third-tier banks wanted to quote their own clients directly and internalize risk. So there were a lot of people trying to quote who didn't really know how, and a lot trying to do volume who didn't understand it. We recognized that's a very good market to be a taker in because you can access liquidity at a fundamentally mispriced price. Our strategies in FX were like 95% taker strategies, just buying the offer as big as possible because that liquidity gap creates the momentum you need to make money. Eventually, banks figured out it's very hard to quote that way and make money. Going to crypto is very different because most people trading are trading their own money, not other people's money. So people are usually a lot more careful. If they have a trade that's losing money, they can't just warehouse it on some weird balance sheet. But on the other side, there's a lot more access to overall flow and information. In FX, that information would sit at a top-tier investment bank's desk; in crypto, you can see that signature flow across a number of venues. So it's a much more level playing field in crypto. We have to work harder at figuring out what's going on, but there's not the tiering where investment banks have a huge percentage of the flow and can move everything.

I think for us, we've always been relatively opportunistic. Initially, we started off just doing basic stuff: buying spots, selling swaps, basis trading, and also rebalancing between Bitcoin, Ether, and dollars, capturing a lot of alpha that way. Over time, we started to have more discretionary models on top where we can identify the signature in crypto, like effects where it'll go up into a mainnet and then dump, or up into an airdrop and then dump. It's about how you access that capital and trade. A good example was Tron in early 2019, where they announced the BitTorrent airdrop and it went 5x, and the backation was insane. We had done good trades with a locked-in borrow of Tron. Our general style is to be as nimble as possible. Since 2014, we've traded only our own money, so we try to be as unconstrained as possible, taking direct principal risks. That means we can make decisions quickly that others might not be able to. That flexibility has helped us a lot. As our balance sheet has grown, we realized the market capacity isn't there for just that trade, so we have excess capital. We do other trades like Grayscale trades, core investing into layer one and application layer. It's like a family office, investing our own money in a Kelly criterion way to maximize expected value given the opportunity sets we see.

That's an interesting question. In some ways, it's less scary because with other people's money, you have to consider reputational risk. When it comes to our own risk, it's easier to define. We silo off our market neutral trading from other types. For market neutral trading, it's very bread and butter but done in a disciplined way. During the March crash, we did very well compared to most firms because we recognize that market structure risk can often be bigger than directional risk. If you have a long-short spread trade on two venues, you're betting that the one you're long on won't liquidate you in a weird way. We've seen a lot of weird stuff in FX over the years, like the Swiss Franc move in 2015 where brokers repapered thousands of trades. From that experience, we recognize that the things that can go wrong are not just about price but also about market function and structure. So we try to hedge those risks and be more aware of them. Being in principle helps us see those risks, making us more comfortable with directional risk, which is plus EV, and more aware of structural risks, which gives us an edge.

The Swiss Franc was pegged to the euro at 1.2 for several years. In January 2015, the Swiss National Bank suddenly announced there was no peg anymore. Euro-Swiss was a very popular resale trade; you'd go long at 1.2, sell at 1.21, scalp that range on huge leverage. Every retail FX broker's book was very long euros. When that happened, the interbank market disappeared because people pulled all their bids. There wasn't much liquidity below 1.2. The market nuked down, and many retail brokers' systems couldn't handle the flow. Clients were on 50x leverage, and there was no liquidity to hedge. There are still ongoing lawsuits today between brokers and investment banks about whether trades at 1.14 were real. The repapering mechanism of FX was based on the idea that reasonable people could come to terms, and it worked until it didn't. Top five dealers on average lost nine figures. One bank internalized the entire market and ended up making over a billion because they bought up everyone, knowing it would trace back. The lowest it went was 0.88 that day. So if you're the only guy who can take risk at that point, you can still make money. From those experiences, we recognize that sometimes you need to ask, 'What do you do if you can't log in for six hours?' Despite all the crazy stuff in crypto, in FX I've actually seen crazier stuff.

Oh, sorry, back to crypto. For us, we prefer to only trade on top-tier exchanges for derivatives, like BitMEX, Deribit, FTX. For spot, we are generally quite conservative on where we trade and execute, just to have a low signature and footprint in the market. Fragmentation is something we're very familiar with, and we operate well in a fragmented environment where people don't know where the flow is coming from. We can execute in a wide range of venues simultaneously.

Yeah, so Serum's approach is very interesting. I think Solana, if it can deliver the throughput needed for a protocol DEX, that can be a pretty big pie because people don't have to KYC; they can just trade from a web wallet. But I don't see it eating up centralized exchanges completely because there are other things you may not want on a blockchain. Users may want their funds in an account with 2FA and a safe experience. There are benefits to centralized exchanges concealing flow. On Ethereum, everyone is doxxed; all the whales and movements are doxxed. Unless you have privacy on that layer, if you're a whale, you don't want to trade spot on a doxxed desk. But if there are ways to conceal that and offer more privacy, that is very powerful. Ultimately, the DEX space is hard to predict. Serum has a good shot because it has FTX expertise behind it. As long as you can bridge to where the assets are and make it usable, there will be demand. Serum's edge is that transaction costs are very low and deposit/withdrawal is cheap. For stuff built on Ethereum without layer 2, you pay $100 to deposit $1,000, losing 10%. That's not a good experience. With Polkadot, Near Protocol, and other side chains, hopefully that will push Ethereum to deliver a better user experience, not just talk about layer 2 but actually deliver it so people can achieve adoption, not just whale games.

Yeah, especially the big ones. Coinbase is often used as a mixer. If people want to send to someone, they'll deposit their ETH into Coinbase and then send it back out because that's way better than linking the two addresses.

You mean just general thoughts on yield farming and DeFi in general?

I think initially, it's a good idea fundamentally to distribute tokens to users because they're the ones who can ultimately improve the protocol, give the right feedback, and have a long-term ownership stake. But the problem is that with incentives, there will always be industrial-scale farming or short-term farming. That's unavoidable, but it's still better than giving distributions directly to investors from a long time ago. So it's not perfect, but a step in the right direction. However, people need to think very hard about what governance tokens look like six months or a year down the line, not just a week. Right now, there's a very short-term focus where people say, 'This is the next fruit or vegetable, we're going to farm it,' so they buy that asset to farm it. It's a bit like 2017 late stages where people know it's a money game or a game of memetics. The reality is that game ends very quickly if a few whales dump. It goes up fast, but it only takes 10% of the supply to think about selling for it to go down 70%. There's a lack of fundamentals in some of the latest stuff. With things like YFI, which had a proper launch with a long-term roadmap, that's great. The debate is still out on which ones will have that community and vision. Sushi is interesting because Chef Nomi was pretty involved. It's a good idea intrinsically, but you have to figure out the roadmap. It's a competitive space. It's tough to say which ones will be around in a year or six months, but some are pretty clear that the founders have no aim of making it last. The cycles are getting shorter. Hotdog comes out and pukes in a day, or Sushi where upon the rebase, the guy nuked the coin into the pool. The cycle is getting shorter because scammers recognize it won't last much longer. Hopefully, it's a healthy reset, making people ask how value actually accrues and whether we can build something longstanding. You can't keep forking everything. At some point, you have to say this is what we use for this AMM, and the distribution we like for this reason. Forking repeatedly and liquidity washing back and forth doesn't accomplish much. It's just a money game, and the winners may not support new games; they may cash out. That leads to an ICO situation where a few people win and take money out, and the whole game nukes. You don't want it to become a poker game instead of building something non-zero sum.

That's a good question. Speaking to some people, they're long-term bullish, but where are we in the latest hype cycle of DeFi farming? If we're not near the top or just come off the top, it's definitely not the beginning. The beginning is marked by people not having made money yet and not really knowing about it. Now, people have seen the YFI wealth creation and want a piece, but the fundamentals aren't there for new ideas. So you end up with a lot of desire to make money but not many ways. As long as everyone believes a coin is worth X, everyone's APR is high. But the moment people realize you still need a buyer for that coin for the APR to be that number, the mirage shrinks and breaks instantly. Once broken, it doesn't come back. In late 2017, early 2018, all the ICOs held their ETH, but as soon as they realized not everyone was holding, they dumped. Once the mirage breaks, the belief that you can hold that coin and still make money in a month is gone. That's honestly healthy because it makes people more skeptical of new farming things. People will see what's already built and ask how to contribute. If Sushi succeeds and becomes Uniswap V2 for a while, people should build on that, contribute to it, make it more useful. They should not fork it again just to farm another thing. The goal of farming is not farming itself; it's to achieve a distribution that incentivizes improving the protocol and distributing ownership. It is not farming for farming's sake. In the past few days, it has become farming for farming's sake, all about showing high APR and making it as bizarre as possible.

I think it just bursts for a segment of the market and makes people question the value of governance more. People will have to prove that their governance has value instead of just assuming it. Right now, people assume governance has value, so if there's X amount farmed, there should be X dollars in the coin being farmed, which is an obscene concept. The money farmed is not necessarily risk or stable. So a lot of these valuation memes will get attacked over time. It doesn't take a big crash; if numbers don't go up anymore, people will question. Especially if layer ones like Bitcoin, Ether, Polkadot, Solana start going up while DeFi coins stay flat, people will ask if these memes hold. The market needs better metrics. I agree that even if most mirages turn out to be mirages, there's a lot of true innovation, and we're long-term bullish on DeFi. The market clearly got ahead of itself and was getting very greedy. The scammers are out, trying to deploy as fast as possible. If it's like spring 2017, it won't matter; they'll do it anyway, and it will grow through it, and some mirages will even go legit. But right now, I'm still mixed on how that will develop. I can see it both ways.

Yeah, there's definitely way less mirages than before. A lot of key myths from 2017 about scalability and what blockchain is useful for have been debunked fairly well. People don't buy into hype the same way. Even with DeFi, at least it has a mainnet; you can interact and see how it works. The code has to actually work. In 2017, stuff didn't even have to work; you just paid for a whitepaper and a consultant to get listed. Even coin distributions were fake; one guy would buy all the coins from a crowd sale and claim it was a thousand addresses. Then he could make a couple hundred x and sell 1% for free money. Those games don't happen anymore because users are smarter, and there's more on-chain forensics. When coins move on chain, everyone talks about it. The IQ level has gone up a lot.

Yeah, for sure. No problem.

I think a good rule of thumb is to generally have a portfolio plan. Have a thesis for why you're joining the space. If you're bullish on Bitcoin long term or Ether long term, have some fixed allocations you believe in and be willing to make or lose money based on those. Don't take all your capital and chase the newest VAD. Make sure you have some long-term holdings. For whatever you trade that's more short-term, have a way to grade yourself on whether you're right or wrong. Falsifiability is one of the most important things in trading, being able to falsify whether you are good or bad at something. Having that segmentation and discipline is key. The joke is that everyone is a trader until they're down on the position, then they're an investor. Don't become an investor in things you were originally trying to be a trader in. Everyone's made that mistake. That's the biggest key to avoiding big impairing losses that are hard to come back from. At the same time, if you're good at something, don't be afraid to do it bigger and try to scale it up. There are a lot of opportunities in crypto, so it's important to take advantage of them while they're there because markets change quickly. If you're good at farming or trading DeFi coins, maximize while you can. If you're good at leverage trading of alts, size it well when you know you're right. Another mistake people make is sizing all their bets the same or sizing their worst bets bigger just to double down and get back to flat. Often, people don't size big enough when they're most ripe; they start taking profits off when it goes in their favor a bit, or if it comes back to break even, they take off. It's all psychological. Always being aware of your conviction level on a trade and being able to quantify that is very important.

I think there are a few things I've learned over the years. One is that there are always opportunities, so don't be afraid to pivot. Don't be afraid to be open-minded and hear what people have to say. A lot of entrepreneurs are scared of alpha leakage or people copying their idea. Don't be afraid of that because they're just as likely to give you your best idea as you are to them. Be confident in your own ability to execute and have that non-zero sum thinking. That's one of the most important things as an entrepreneur. You have to be able to attract talent and be confident in your own talent to execute in a competitive world. In trading, some things are more zero sum and you have to be more secretive, but as an entrepreneur, you have to convert people to your vision and what you're doing.

Thanks for having me. I had a lot of fun.