How to Spot Trading Opportunities | Don Wilson, 1000x

I moved back to Chicago in 1994, and I said, “If this stuff—if the CME—hasn’t completely transitioned to the screen by the year 2000, this place is going to be out of business.” I just couldn’t have been more wrong about the timing. It took far longer than that. I think these transitions always do.

This time, we have a very special guest who has not done a lot of podcasting before, and we’re very lucky to have him on. He’s the jefe of Jonah, my partner in crime on this podcast: Don Wilson, the founder of DRW. I’ll let Jonah, who knows him much better than I do from working together, give the introduction.

Thanks, Avi, and thanks, Don, for joining us. We’ve got a lot of really interesting content to put out today. Don graduated from the University of Chicago and went straight into trading, right into the pit in Chicago. In 1992, he founded DRW, which is a global trading firm that touches pretty much every asset class, with both proprietary trading and liquidity provision.

It’s a pretty exciting company at the intersection of trading and technology. Don, thank you very much for joining us. Without further ado, can you tell us what it was that drew you to trading straight out of college? How did you know that you wanted to make that your career?

When I came to college, I had no idea what I wanted to do, but I knew that I was good at math and that I liked using math to solve problems. My initial inclination was that maybe I’d do something in science. As I experienced college and the academic setting, I was concerned about the very slow feedback loop that happens in academia.

I raced sailboats in high school and college, and one of the guys I raced with was working on his postdoc in physics. He eventually went on to fly the space shuttle, but at the time, in addition to working on his postdoc, he was fiddling around with FX modeling. I thought, “That’s really cool. You get this immediate feedback loop, and you don’t have to write a paper about it and present it and not really know if you’re right or wrong. You just figure something out, go see if it works, and find out right away.”

That sounded like a lot of fun, so that’s when I decided that I wanted to go into trading. Obviously, being in Chicago, the trading pits were right there.

What was it like when you first stepped into the pit? What was the process of getting up to speed and figuring out how to trade? I can imagine that you walk in there and it’s hectic and crazy. How did you figure it all out, and what did you start with?

I graduated early, when I was 20. I was an extreme introvert and had a real hard time interacting with people, but I got a job with a small trading firm called LETCO. They said, “We’ll have you spend a couple of weeks standing next to different traders who work for us.” After 6 months, they said, “Here’s $100,000. Good luck.”

I became a member of the Chicago Mercantile Exchange, leased a seat, and decided to go into the Eurodollar option pit, which was 3-month LIBOR options. At that point, you walk into the pit—the pit opens at 7:20 in the morning and is open from 7:20 to 2:00—and find a spot to stand.

It’s pretty intimidating because it’s super crowded, with a bunch of people pushing and shoving, yelling and screaming. It looks like total chaos, but the reality is that it’s just math. I’d stand in the pit during the day, then go home and write code, build option-pricing models, write risk software, and fiddle around with different volatility models, trying to model volatility surfaces.

Then I’d go back the next day and see if any of it worked. After a little while, I started to figure some things out. My whole mindset was: How do I define the fair value of every single option across the whole surface? Then how do I use that framework, along with the order flow coming into the pit, to pick up small amounts of positive expected value?

That’s how it all started.

What’s an example of something that you went home and coded, brought back into the pit the next day, and saw work?

There were a few iterations that were helpful. First of all, I had some stuff that was not all that innovative, but I designed my own sheets with all my own option values on them. I actually used an ImageWriter II to print them, so they took hours to print. I lived in a studio apartment, so I had to put a pillow on top of the printer when I went to bed.

I’m not sure all that stuff was amazing, but it was nice to be able to configure everything, have calls and puts in different colors, and make those incremental improvements. The risk views that I developed were really pretty good. All the different scenarios and those formats were certainly much better than the off-the-shelf vendor tools that you could buy.

The places where I started to pick up valuation edges were when I started thinking about how the volatility surface evolves in the fixed-income space. A couple of years after I started, they listed serial options—short-dated options on longer futures. In this case, they were July and August options on September futures, so you’d have July, August, and September.

In the fixed-income space, generally, the front of the curve is less volatile than further out. At the time, typically, it was the first red future—the future that was just over a year out—that was the most volatile. Lately, of course, it’s been further out the curve. During the period of low rates, it was much further out the curve, and more recently it’s moved toward the front of the curve.

That peak of volatility evolves depending on the Fed regime, but generally, almost always, the front of the curve is less volatile because the Fed controls the overnight rate. That’s relatively static unless the Fed changes it. There was this persistent volatility roll-down where the front month was less volatile than the second month, the second month was less volatile than the third month, and the third month was less volatile than the fourth month.

When they listed these serial options, the natural assumption was, “Obviously, if June volatility is 20 and September volatility is 30, then July and August have to be somewhere in between.” But that’s completely wrong, because July is alive when the September futures are more volatile than that forward period after July expires.

If you think about those volatility forwards, July should obviously trade at a much higher volatility than September, and August should trade somewhere in between July and September, but certainly well above September as well. That was an early realization I had in modeling this stuff, and the market did not understand it that way at all.

That was definitely the kind of thing that gave me an edge because I had a very clear idea of what the fair value of these things was, while the market had a very different idea. They were pretty low-variance trades, so I traded a lot of those types of things.

Your trading style must have evolved over time, especially as we transitioned from pit trading to electronic trading. That transition was not managed well by a lot of people, with you being a major exception. You managed that transition exceptionally well. What was that like? Why do you think you were able to make that transition effectively? And does coming from the pits give you an edge today?

One of the things about the trading pits is that it’s ingrained in you that you need to be able to price anything at any time, at a moment’s notice. If you can do it in your head, you have that much more of an edge, and I think that practice is still useful to this day.

When September SOFR futures had a 100-plus-basis-point range in March, just thinking about the volatility surface in that environment and having that really intuitive grasp of what was going on, I think, gives you a perpetual advantage.

Shortly after I started, they launched Globex, which was an electronic trading system. I had a pretty small operation at the time, but I thought, “I definitely want to get a Globex machine. I definitely want to have someone making markets overnight,” because that’s how this stuff should trade. I was excited about it when it first happened.

I still remember that Globex’s software and hardware were developed by Reuters, and the CME licensed it. They handed out T-shirts on the first day that Globex launched. On the front, it said, “CME by day,” with a big sun. On the back, it said, “Globex by night,” with a moon.

I thought it was an awesome lie, obviously designed to make the pit traders feel comfortable that their jobs weren’t at risk.

In 1992, I moved to London and set up the London operation. I stood in the Bund option pit for the first year, then traded from upstairs for the second year. That was right when Eurex fought really hard to get the Bund futures to transition from LIFFE to Eurex.

We had a team trading the arbitrage between Eurex and LIFFE. We had our own broker in the pit in the Bund futures pit, and sometimes our broker would go out for lunch, have a pint, and come back a little bit off. I thought, “This pit stuff is just ridiculous. The sooner this stuff transitions to a computer, the better.”

Then I moved back to Chicago in 1994 and said, “If the CME hasn’t completely transitioned to the screen by the year 2000, this place is going to be out of business.” I just couldn’t have been more wrong about the timing. It took far longer than that.

I was always a big believer that this stuff could be handled more efficiently electronically, but I was definitely dead wrong about the timing.

That probably meant you had more of an edge for longer, given that the transition took longer than you expected. It was definitely a boon in the beginning of DRW.

You said you have to be ready to price anything at any time. There’s a price for anything—the classic pit-trader mentality. What was it like when a whole block of this new thing called Bitcoin came up for auction sometime in 2013 or 2014? You had to think through pricing it and maybe even starting a company around it. How did that work?

We had a number of people at DRW who were interested in Bitcoin. I started thinking about it, read the paper, and was super intrigued by it. The idea that you could transfer value in a trustless manner was very appealing to me. The lack of intermediaries can dramatically reduce friction.

I had a very dim view of intermediaries because, over time, I became the largest trader in the Eurodollar option pit. Often, trades would come through the pit, and we would take down a huge chunk of one. Then 10 minutes later, the block trade would go up on the board. You’d think, “Some big bank just ripped off its customer and then came and back-to-backed the trade into the pit.” That’s not even trading. It’s ridiculous.

Intermediaries were not providing a ton of value in the system. I’m saying this in the nicest possible way; at the time, I wasn’t saying it in such a nice way.

The idea of a financial system that reduced intermediaries was naturally appealing to me. We had debates about whether this was important and what was important about it—whether it was Bitcoin or the blockchain. All the debates that people eventually started having, we were having in 2013.

We decided to do a few things. We started Cumberland to provide liquidity in Bitcoin. We started a company called Digital Asset Holdings, based in New York, that was more focused on blockchain applications. We also decided to buy some Bitcoin.

Then the government auctions came up. A lot of those were Silk Road coins that had been seized. At the time, there was a misperception that criminals were going to use this stuff to do bad things. The advantage for criminals, I guess, is that they don’t have to go out and pick up a sack of $100 bills. The bad thing for criminals is that every single transaction is memorialized forever.

That doesn’t seem like a good characteristic to have if you’re going to try to do something illegal. Silk Road found out the hard way that it was a really dumb thing to do.

That’s actually the reason we named Cumberland Cumberland. We figured that, at the time, people believed this was only going to be a place for criminals, so we should at least distance the brand a little bit from the core company.

The auctions started happening, and to us it was just a matter of understanding the liquidity in the market and pricing the auctions accordingly.

When you first spun out Cumberland, where did the name come from?

One of my longtime partners at DRW, Jeff Lough, was a lawyer in his former life, so he was often involved in naming new companies. Jeff is a big Grateful Dead fan, and there’s a mining song about the Cumberland Blues. Given the mining link in cryptocurrencies, he decided to call it Cumberland.

It’s a good story.

I understand why, in the beginning, you wanted to separate it out. When you were first spinning it out and thinking about what it could be, what were you envisioning Cumberland doing or providing in the space? Was it always going to provide liquidity, or were you thinking of it more as a risk-taking venture? What was the thought process behind starting it?

The idea was simply, “At DRW, we know how to trade, we know how to take risk, and we know how to provide liquidity to the market, so let’s do the same thing in Bitcoin.” That was the genesis of Cumberland.

It was much more of a counterparty-facing situation, but we were also providing liquidity and trading on other venues.

You’ve obviously been through multiple crypto cycles. More broadly, you’ve been through cycles of market maturation. You were talking about Globex and things going to the screen. Is there a recipe for surviving cycles that you’d point to? You’ve managed to do this through so many different peaks and troughs of euphoria and despair—not just in crypto, but in lots of different markets. How do you think about navigating these things?

There are 2 different cycles that you’re talking about. One is a market-structure cycle, and the other is a hype cycle, or bubbles.

The way I think about it is that I think about risk management first and foremost. It really all comes down to basic risk management.

I will say that Bitcoin and cryptocurrencies in general, but especially Bitcoin, are some of the hardest instruments to trade. The only thing that is clear is that, when there’s a lot of hype, that increases adoption. That increases the probability that Bitcoin becomes viewed by a majority of the world’s population as a superior version of a store of value.

On the other hand, when the price of Bitcoin declines, people become less interested, and the perceived probability that Bitcoin becomes a replacement for gold declines. Both of these things are important.

When the price is going up, you can build a model that argues that the fair value is higher than it is now. Conversely, when the price is going down, you can build a model that argues that the fair value is lower. I think it’s unique among financial instruments to have that characteristic.

In commodities, you have almost the opposite. When the price goes up too much, people demand less, and then the price may reverse. But if Bitcoin is a negatively convex asset that creates self-fulfilling rallies and sell-offs, what makes it harder to trade? Wouldn’t you be able to implement a momentum strategy that follows the trend, where there would theoretically be large moves?

Maybe it’s a momentum strategy, but it has big troughs and valleys along the way. It’s very easy, if you don’t have conviction, to get washed out.

If your strategy is levered long and you say, “If it starts to go down, I’m going to get out,” it’s probably a money-losing proposition.

Are there any other markets that you’ve seen trade similarly to this? What is the most similar market you’ve seen, if any, to crypto?

I think it’s really a very different market. The closest thing is that you see similar price patterns in emerging technologies, like the internet bubble and that kind of thing.

I would argue that those things are fundamentally different because you’re still talking about the valuation of a company. You ultimately have to make assumptions about that company’s earnings, market share, and total addressable market.

With Bitcoin, it’s not an earnings question. It’s simply a question of whether this is a superior store of value to gold. That’s really a question of what percentage of the world’s population feels that way.

The answer to that question also seems a little bit path-dependent. The more people who adopt it in a shorter period of time, the crazier the momentum and the greater the outperformance. It’s a very interesting asset to trade.

One thing that you said is that emerging technology sometimes acts like crypto. We’ve been hit with this recent wave of AI applications and large language models. At least in the short term, over the last 6 or 7 months since ChatGPT really went viral, there’s been a lot of discussion about how it might impact our jobs and the financial markets.

As someone who has navigated a lot of change in technology and how it impacts financial markets, how do you see this recent development? Are you keeping your eyes on it?

Right now, I’m in Woodside, California, 10 minutes from Stanford. I’ve met with a bunch of AI people over the last few days. I’ve been coming here regularly and doing that.

I think it’s a consequential technology. The innovation taking place in large language models specifically is really very interesting. I think it will impact pretty much everything, aside from somebody living off the grid in a cabin in the woods.

Is this one of those moments in financial history when an enterprising young person can go back to a studio apartment, use some new technology to print out some sheets, come back to the pit the next day, and win? Is an LLM the modern equivalent of that, in that it will give the first group of traders who figure it out an edge? If so, how do you think about that?

It’s entirely possible. I’m convinced that LLMs are going to be really useful for thinking about some aspects of markets and risk-taking, but it’s pretty unclear exactly what those are and how much human intervention you need in that process.

It’s a matter of discovering that very shortly. Then the technology will continue to evolve, so the answer to those questions will almost certainly change.

It’s interesting to think about whether we’re at this Deep Blue moment, where technology is about to beat human traders at their own game, or whether we’re more at the CME-going-to-the-screen moment, when it should have happened years ago but still hasn’t.

Now it has, but maybe in 2000.

Has AI changed the way you run your life or your business, or changed your approach to markets? Or are you still in information-absorption mode?

Right now, I’m in information-absorption mode. I wouldn’t say that I’ve changed anything meaningful other than spending time on it.

Please let us know when you do know.

I was reading some of the articles you’d been mentioned in before doing the podcast, and I found a funny quote. When the LME had its nickel incident, you called it “one of the most inept moves I’ve ever seen an exchange make.” I thought that was uniquely funny for someone who has spent so much time in crypto, because we’ve obviously had our fair share of dealing with venues.

You’ve seen the crypto market evolve in terms of liquidity over time. How do you view the last 8 or 9 years since you’ve been in crypto, how has it evolved, and where do you think it’s going over the next 5 years?

There’s obviously a lot less excitement about it from outside the crypto community than there was before FTX blew up. But I still think there’s really important innovation taking place in the space.

The ability to move value instantaneously and, if you want, trustlessly is a really important innovation. I think leveraging that technology for traditional financial markets is one of the most exciting and impactful ways this technology will be used. I’m excited to see that unfold.

In terms of how we interact with the markets, there’s currently a separation in the traditional world between where liquidity lives, who can custody the asset, who can provide leverage, the exchange that provides it, the brokers that provide it, and the prime brokers that provide leverage.

Do you think that’s where crypto will eventually go? Is that the endpoint of financial markets, or is crypto going to show us a different way that financial markets could operate?

In the centralized crypto-exchange space, these exchanges decide that they should do everything. They should be the FCM, the DCM, and the DCO, and then, in addition to that, they should provide leverage.

As I was watching this unfold, I thought, “This is inherently less stable.” One of the nice things about traditional financial markets is that, by breaking up these different responsibilities, you get more transparency, safety, and resilience.

When you switch over to the DeFi world, you can open up those different responsibilities to different market participants in an even more granular way than you can in traditional finance. It also opens up the ability to provide short-term loans and move money instantaneously.

The LME decided not to ask for margin because it was going to blow up some of the large nickel producers. Then, of course, they decided to cancel all the trades one day because they were worried that a bunch of members of the clearing house and customers would default.

That’s one way of running things. The other way is to say, “We’re going to move value in real time. When there’s a margin deficit, you top it up right now.” You have the ability to do that because you can move money on-chain.

My belief is that this actually leads to a more transparent, safer, more resilient financial market. That is the promise of a lot of this technology, which will enable that to happen.

What sort of time frame would you predict for the blockchainification of certain assets—perhaps nickel, or things that right now aren’t necessarily associated with crypto?

I’m perpetually wrong in estimates like that, so I always think they’re going to happen sooner than they will. Whatever I say, just take the over.

There are some assets that are physical assets, like nickel. You can put nickel on a blockchain, but ultimately the nickel has to sit somewhere. If somebody steals the nickel, then you have a blockchain that represents nickel, but the nickel isn’t there. That’s a problem.

That kind of asset still has this intimate link to the physical world that is supercritical. Other instruments are already virtual instruments, like equities and Treasuries. People don’t usually walk around with their share certificates, and I think those assets probably lend themselves more to digitization and blockchainization.

That’s going to happen. Right now, we’re experimenting with intraday repo using blockchain technology. That’s built on Canton, which is Digital Asset Holdings’ blockchain, and it uses DAML, the smart-contract language on top of it. It’s powered by Broadridge.

That’s an example of experimenting with the early days of using this technology. Ultimately, it enables value to be moved in real time, even 24/7. That’s the kind of thing that can make clearing houses more resilient if they choose to avail themselves of that technology.

Could you tell the listeners a bit more about Digital Asset Holdings, the Canton blockchain, and what sort of problems it solves that Ethereum doesn’t?

Ethereum is obviously a very powerful network, with a tremendous amount of energy and participation. One of the drawbacks of Ethereum is that every single transaction is public.

For most people, if they buy or sell a security, they don’t want the whole world to know that they bought or sold it. It may be fine for their counterparty to know, and perhaps for a regulator, clearing house, or prime broker to know, but for any given instrument and transaction, you want to be able to control who can see it.

The advantage of Canton, and of the smart-contract language on top of it called DAML, is that it has configurable privacy built into the core of the chain. The team has been working on this since 2014, so it has been a very long process to build a blockchain with those characteristics.

What is the issue with the current stack that prevents this from happening? With the current architecture of traditional financial markets, why can’t we achieve what Canton enables?

We understand the value that blockchain can bring to the financial markets, but it would be useful for people who are less familiar with it to understand the current issues.

If you want to wire money right now, that can take hours, and you can only do it when the wire windows are open.

Let’s say you have a spread between a futures contract in London and a futures contract in the United States. After London hours, there’s a big rally, and you’re long the futures in London and short the futures in the United States.

The idea that you could pull variation margin out of the London market and move it over to satisfy the negative variation margin you’re experiencing in the United States is not even close. What will happen is that the next day, assuming the market opens unchanged, the London futures contract will rally and the variation margin will show up in your account.

At that point, you can wire it out—maybe that day, maybe the next day. Maybe you’re now running into the weekend. The whole system is very clunky. A lot of it is done in batches, with cycles that process things once or twice a day.

It makes everything very slow to move around and means that you need to have a lot of extra capital in the system to deal with all those delays.

That’s very helpful framing for people. We’re participants in the crypto market, and we’re honestly blessed with a system that is accessible 24/7. Every time I go back to the traditional markets, I leave feeling a little disappointed with what I can and can’t do. Then I flee back into my MetaMask and end up much happier.

It’s good to hear that these problems are being tackled. It’s something we’re watching closely as well.

One last question for you, Don. When people think of the Chicago trading firms, a lot of them think of high-frequency trading, taking a few basis points here and there. A lot of it is arbitrage, and I think much of your career has been characterized by finding those types of arbitrage opportunities.

But you also take risk. I’d love to hear about your process for decision-making when it comes to taking risk, and what you think has helped you the most in your career when it comes to being right.

The little arbitrages are great because you can generate these very high-Sharpe, high-return trades. The reality is that they don’t persist. Markets constantly become more efficient, and those opportunities quickly disappear.

Building a business that relies only on capturing those little arbitrages isn’t sustainable. Being comfortable moving out the risk curve is super important. If you can move out the risk curve while having access to very low-latency tools and connectivity, then you’re in a really strong position. That’s where we try to be.

When I think about risk—when I think about markets, or really most things in the world—I see a series of probability distributions. If you look at the world as a series of probability distributions, you’re well set up to think about risk.

Any time there’s some perceived dislocation in the market, or a violent move, there’s a high probability that something is mispriced. It was either mispriced before or it’s mispriced now. There may be some fundamental thing that changed, but there’s still a high probability that it’s mispriced because new information has come out and there’s been a dramatic move.

I think about the drivers of the price action and the drivers of supply and demand. If it’s a commodity, I think about those drivers. I think about the Fed’s reaction function, as well as the supply and demand for the hedging needs of that part of the interest-rate curve.

Then I try to think about what could change going forward that would shift the Fed’s reaction function one way or another. What would cause the Fed to ease right after Silicon Valley Bank?

The interest-rate market priced in an expectation that the Fed was going to ease by 100 basis points by the end of the year. Of course, that’s not what happened. So far, the Fed has hiked 50 basis points since then, and I expect the Fed may well hike again before the end of the year. I expect the Fed won’t ease by the end of the year.

Those outcomes were all within the probability distribution, but my view at the time was that the probability of the Fed easing by the end of the year was massively overpriced. That part of the distribution was overpriced, while the possibility that the Fed would continue to hike was significantly underpriced.

That makes a lot of sense. I always try to think similarly. One of the hardest parts is constructing that probability distribution—thinking about which criteria actually go into building it. You say, “I think there’s a 60% chance of this, 30% of that,” and then it turns out you were completely wrong.

Coming up with your own subjective measures of a probability distribution is super hard, especially when the market is saying something very different.

As a closing question, are there any trades that you really loved that you can talk about? Maybe your favorite trade ever, if you’re willing to disclose it. Is there one you can think back on and say, “I really loved putting this one on,” whether or not it made you the most money?

I’ve done lots of trades that I really loved. This year, I will say that the dislocation I perceived in the interest-rate market after Silicon Valley Bank was one of my favorite trades.

I thought there was such a gap between my perception of what the probability distribution should be and the market’s implied probability distribution. I thought that was phenomenal.

Going back in time, one of the trades that comes to mind is from 2008, when Lehman went bankrupt. We were one of a handful of firms that the CME called to ask to price its portfolio.

We were well prepared for it. We got the whole portfolio from the CME, broke it down into different chunks of risk, had different teams price the risk, added it up, and submitted the aggregate bid. It was actually by product, but we ended up being the best bid in 3 of the 5 buckets in the auction.

When we won those buckets, we were very efficient about hedging off that risk in the optimal way. The entire process took a lot of confidence in how we were thinking about and modeling the risk, as well as the ability to aggregate it and, in a very volatile environment, say, “Here’s our number.”

That’s an awesome story. It’s part of history.

I was part of the Lehman portfolio, but I didn’t get acquired. I had to go find a different job after that.

What happened there?

I had to go find a different job.

I had one follow-up question for you, Don. Does your Bitcoin trade factor into your top 5 trades, or did that feel more like a venture investment than a trade?

That was a series of auctions, and they were relatively small in terms of risk. They were interesting because we had never interacted with the United States Marshals before, but I thought they did a good job thinking about how to maximize the value.

We were very comfortable trading the asset class, so it was great.