Decisions, Decisions, Decisions: An interview with Todd Simkin

On August 2, 2023, Kevin D of Quantitative Traitor interviewed Todd Simkin, a director at Susquehanna International Group (SIG). Over the last 20+ years, Todd has traded equity options, ADRs, currencies and fixed income products. Currently Todd is CEO of SIG Re, a Bermuda-based reinsurance company, and an investor with SIG’s private equity team. What follows is a lightly edited transcript of the interview.

Kevin D: SIG trades in all sorts of asset classes today, but it got its start in derivatives — equity options and then index products. How did that come to be?

Todd Simkin: Most of the founders of the firm each had their own company before coming together to form SIG, and they were all on the floor of the Philadelphia Stock Exchange trading equity options. The businesses grew, particularly Jeff Yass’s, and he also started trading index option products, which were particularly interesting at the time because there was no great way to trade the underlyings. There were no ETFs, so no way to get in one equity trade exposure to the broad index. You instead had to trade baskets of stocks appropriately weighted by the number of shares represented in each index, which made the calculation of their hedge harder and the calculation of the volatility of the underlying index more complex than it is today.

As the businesses grew, they decided that it made a whole lot more sense for them to join forces and work together in their trading instead of merely staying out of each other’s way. In 1987, they came together to form Susquehanna Investment Group — the original “I” in SIG was “Investment”, now it’s “International”. We started trading together, and the other products we traded over the years were natural outgrowths of our expertise, the order flow and opportunities we saw in the market.

KD: How was your experience starting out at SIG, trading equity options on the floor of the Philadelphia Stock Exchange?

TS: I started trading at a time when stocks and options were still traded in fractions. A stock would’ve been quoted as 64⅛ to 64¼ instead of 64.125 to 64.25. That makes me a dinosaur.

When I started trading, I went through SIG’s education program and learned how we approach the markets and think about risk, risk pricing, the shapes of distributions and everything that goes into option pricing.

Then I returned to the floor to trade options on Dell Computer. It was the most widely traded equity option at the time, and it was one of the first options that faced multiple listings. Options originally were traded only in one location. If you wanted to trade options on Motorola, 3M and Bank of America, you had to trade each one respectively in San Francisco, Chicago and Philadelphia. Dell at the time was only traded in Philadelphia, but as I joined the crowd, the rules changed on the exchanges, and now you could multi-list — options on one product traded in multiple locations. Dell was now traded in New York, Philadelphia and Chicago, Amazon was traded in Chicago and Philadelphia. All of a sudden, there was a lot more competition and difficulty in tracking the top of the order book — the highest bid and lowest offer — in all these names. One of the challenges I and everyone else faced at the same time was learning how to trade multi-listed options.

I also faced the natural educational challenge of learning how to trade altogether. How to stand in the crowd and be noticed by brokers. How to make two-sided markets. How to avoid losing money by setting up arbitrages or missing relevant information for the pricing of the securities you were trading. How to take advantage of opportunities presenting themselves — looking for parts of the distribution priced out of line with the rest, where you took a position that would hopefully be profitable in the long run.

KD: How was going through SIG’s education program?

TS: Our training program has remained pretty consistent throughout. We’ve certainly changed our approach to teaching as technology and the marketplace have changed. But ultimately we teach the basics of derivative valuation and option pricing through the lens of uncertainty and imperfect information. We teach the math behind pricing models, starting with Black-Scholes and then moving onto adjustments. Zero people in the marketplace today use Black-Scholes out-of-the-box. Everybody makes their own adjustments — adjusting volatility smiles/curves/surfaces is first-order, adjusting skewness and kurtosis is higher-order.

Learning how those models worked is the first step. Once we understand option pricing, we look at the relative value of options to each other — not only theoretically, but also practically through mock trading. We make and talk through the same decisions we’d make with our own capital, but with fake ledgers and money instead. All this is done in the context of behavioral economics — understanding and making sure we’re aware of the biases and mistakes we and others have and make and express in the marketplace, so we can capitalize on those of others while avoiding our own ones creeping into the trading process.

KD: You moved on to trade American Depositary Receipts. How did that differ from trading equity options?

TS: At first, it felt like none of my background would be relevant because the underlying product was completely different from equity options — the arbitrage relationships I learned didn’t apply. I had to learn how currencies traded. This was pre-euro, so I was trading multiple currencies — Deutsche marks, French francs, Mexican pesos against US dollar positions. As I was doing that, the part of my education that continued to matter was thinking about making decisions with imperfect information, biases expressed by ourselves and others in the marketplace. All that was transferable knowledge and information to trading ADRs.

But learning operationally how these trades worked was very different. The ADR trade was trading a common stock in a foreign country and then trading that same stock on US exchanges. You bought the stock in Deutsche marks and sold it in dollars, hence needing to do an offsetting currency trade as well — selling off your dollars and buying back your Deutsche marks. There was a three-legged arbitrage I had to get a good handle on. Ultimately the most profitability that we saw in these markets didn’t come from understanding the arbitrage, which was easy, but came from figuring out the deeper game theory on when and how to take positions to maximize the information we interpreted from the marketplace.

KD: You moved to Europe for a while to set up SIG’s operations in Dublin, Ireland. What was it like moving from trading to setting up SIG’s operations in a new country?

TS: It was fantastic to be there literally from the ground up. My first day on Irish soil, I was wearing a hardhat and looking at new construction spaces to figure out where our office would be. I went from everything up until then being operationally taken care of for me on the trading side of the business, to the team teaching me tons about the technology, compliance, operational, legal sides — it was an on-the-fly education on how to set up a business. One day I was finding office space, the next day I was working with our European brokers to get set up and run test trades to ensure we could clear trades appropriately across different accounts.

KD: You moved back to the US from Ireland to work on business administration and development. How did that compare with working on the trading floor?

TS: I was brought back to work directly with our Chief Operating Officer — one of the firm’s founders — who needed help running the day-to-day business. I transitioned from thinking about trading decisions to business decisions — how to compensate our traders, how to set up an HR group and a recruiting team. It was very important for us that people in the business roles were traders — people who not only understood P&L, but also the impact of luck on outcomes, how to appropriately compensate for variance, how to talk to our prime brokers about leverage and borrowing.

Having set up an operation was a big help, but working directly with somebody who had been doing it for years was also incredibly helpful. Speaking the language of traders meant that we talked about business decisions in the context of trading risk as opposed to how a business school might teach students.

KD: In 2007, you moved back to trading to trade fixed income. What had changed about trading in the interim?

TS: In the interim, everything had become more electronic, which led to more transparency in the marketplace and most decisions being more highly levered. It became easier to have a broader impact with the same trade than when you had to execute it on the exchange floor and match up a paper ticket to another paper ticket. A good amount of additional complexity came with the move to electronic markets.

In the interim, I was following all the developments in the market, and I helped start several new businesses for SIG in that time — our investment banking business, our private equity and venture capital arms, our energy trading business.

I moved back to trading in 2007 because we needed another senior set of eyes in the fixed income business — it was around the time of the financial market meltdown, where Bear Stearns and Lehman Brothers failed. The financial crisis and the lead-up to it were a story about a disjunction in the fixed income markets. I moved there to work with our team of fixed income traders to help manage that considerable risk — we needed operational know-how in addition to trading know-how.

KD: You mentioned everything becoming more electronic in the interim. How did SIG transition from trading by hand to trading by computer? 

TS: It wasn’t something we were doing outside of what was happening in the market. We always had computers behind-the-scenes doing some of our work. Even when we were trading by hand, we’d run models on computers to run printouts at different prices and levels given different assumptions, and then we were flipping between sheets in our hands and making adjustments given then current market conditions.

When we moved to computers, instead of running those models in the background at the beginning of the day and then printing out sheets, we ran them continuously through the day, driving the pricing that was tied to the market.

The big difference we saw was an exponential rise in the number of technologists that we had. When I started off in the mid to late 1990s, our technologists’ day-to-day job was making sure the computers turned on and were connected to the market. The traders then ran theoretical values at the beginning of the day and updated once or twice as the day went on. The technologists represented a small fraction of our workforce.

As the markets became more technologically driven, our technologists went from being a tiny minority to a majority. There’s so much more to be done with processing the vast amounts of information coming in and then ensuring we’re as fast as possible in connecting to the markets, getting our markets out there and driving the automated responses and changes we want — all happening faster than human time. 

KD: After a few years trading fixed income you moved to the education side of the business in teaching new traders. One thing you do with new traders is getting them to articulate why they traded the way they did, no matter if their trade was good or bad, won or lost money. How has teaching this practice evolved over time?

TS: The practice of talking about the factors that went into a decision process irregardless of outcome has been consistent from when the company was founded to when I was teaching and to today, where we’ve got other fantastic traders running our educational program.

What’s changed are the analytical tools at our fingertips. When SIG was founded, we didn’t have the real-time analytics of options pricing that we have today. We didn’t even have access to the information that we’d want in order to investigate correlations between stocks within the same sector, measures of risk given certain counterparties, among many other things. As we developed tools internally and rolled them out to our trading team, we also rolled them out in the education process to teach our younger traders how to use and implement them to fine-tune the implied distributions of markets they were sending out to the world.

KD: How do you get students who’ve studied math and know Bayes’ theorem to make asset allocation decisions with imperfect information?

TS: Understanding Bayesian thinking is necessary, but it’s not sufficient if you don’t know what information you’re using to update and how to weigh it. A big part of our training program is teaching by example, where we’re creating scenarios for the students to respond to and then talking about what information is relevant in the scenarios and how to incorporate it in that Bayesian way. Any high schooler can handle the math of Bayes’ theorem, but figuring out when there’s relevant information in order to update your priors and how to weigh it is way harder. That’s our secret sauce — we excel in teaching people what information is relevant and how to incorporate it in that process.

KD: How does teaching finance to new traders at SIG differ from teaching it to students in a university setting?

TS: Some of the fundamentals will be exactly the same. I’ve taken derivatives classes at Wharton, and there are things we teach that are identical to what they teach. When we’re teaching the Black-Scholes formula, it’s not like we’ve got secret variables that Black or Scholes were unaware of.

Things differ when we implement this in trading. I mentioned earlier that nobody trades off of naive Black-Scholes. We at SIG make adjustments, the specifics of which I won’t get into because that’s part of our strategic advantage. But figuring out how to look at the world a bit differently is an important part of what we’re teaching that’s different. We have a different approach to modeling the underlying pricing and risk.

The other thing that differentiates our teaching from classroom teaching is us putting capital at risk behind it. There are lots of simplifying assumptions made when you teach about trading in the classroom. Continuous time, homogeneity of counterparties, equal quality of information on trades in the market — none of these are true in reality. When learning about projectile motion in an introductory physics class, you’re told to ignore air resistance in the model. But if you’re actually trying to model a projectile, you need to then come back and incorporate air resistance in your problem. Analogously, we can’t assume certain things away when we’re facing counterparties in the market.

KD: Tell me about the importance of the pre-trading day experience for the assistant traders. 

TS: Trading is ultimately trying to predict the future, which is difficult. If you knew the exact state of the future, then you would know exactly what trades to make and avoid today. What we’re doing in trying to get ready for that future prediction is getting a good understanding of the current state of the world, which comes from multiple factors.

One important factor is on a macro level, what’s changed and different in the world. The financial condition of a company you’re trading will clearly impact the future prospects of that one company. However, you could look at something broader like the health of the US economy, which will impact all US-listed companies.

You could look even more broadly at what’s at happening globally — how will the financial markets be impacted the Russo-Ukrainian War, for example? That could be everything from interest rates to specific companies involved in war efforts.

Getting a handle on everything that has changed about the world will impact how you forecast future events, which is ultimately what we’re doing in our trading. That’s the first step, and then communicating that effectively and significantly to the risk allocators, the traders, is the next step. Researching to understand what’s happening and then translating that into actionable approaches to the market is the derivative step of having a better understanding of the world.

KD: And now tell me about the importance of the trading day experience for the assistant traders.

TS: It’s a hybrid of the preparation that happened before the day started and real-time response. Part of that preparation comes in some scenario planning. You might know some of the things that could change over the course of the day and have different scenarios ready to go given certain contingencies.

If you know that housing starts will come out at 2:15 PM and you’re trading a home-builder company, then housing starts will be a significant number. You can say if housing starts are above this level, then we want to buy volatility and deltas in this underlying. If they’re above this level but below that level, then we want to be flat. If they’re below this level, we want to buy volatility and sell deltas.

You have different scenarios ready to go, and then as soon as the number comes out, as soon as you have a new clarity on the state of the world, you’re ready to implement that. That’s part of what happens in the pre-trading phase, which allows you to then react appropriately during the trading day. 

But then there can also be unforeseen noise in the market. Using my home-builder example, let’s say they announced that the CEO is stepping down — you have to figure out on-the-fly what to do with that information. Or it could be that you find out there’s a counterparty wanting to buy a lot of downside puts three months out in that home-builder. You don’t know why, but you have to figure out how to respond to that — in a Bayesian way, how to update your priors given this person’s presence in the market and what they want to do.

The trading day is a time of distilling this uncertainty as conditions change. Some uncertainty is clarified in that time, some new uncertainty is introduced.

KD: Your former colleague Michael Steiner told me about SIG’s mock trading back when he led it. How has SIG’s mock trading evolved in the interim?

TS: We start everybody up by making sure there’s the same underlying understanding of some arbitrage relationships and how to quote in the marketplace. We do that through open outcry trading a la 30-40 years ago.

That’s certainly not where we stopped, though. We move from that to machines — mock trading in the same setting that these people will actually trade in. So using our trading systems, but having the backend attached to fake exchanges we’ve built internally to mirror the experience of trading on a real one. Over time, we’ve rolled out more of the advanced tools we've developed so that as the students learn more about what factors are important in the decision process, they incorporate the tools into their pricing or risk metrics, just like the traders do. As we figured out that there was more/better/different information that could be gleaned from the market and incorporated into a decision, we built tools to capture that.

KD: SIG teaches its interns and new traders how to play poker. What about trading does poker teach?

TS: The underlying relationships between risk, probability and outcomes is one of the big ones.

Managing your emotions is another. There are times when you’ll make good decisions and lose, there are times when you’ll make bad decisions and win. An important part of trading is not repeating those bad decisions later despite them working out before and not getting discouraged from your good decisions despite them not working out.

At the poker table, we talk about what information was available at the time you made your decision, what decision you made, and in hindsight, was that the best decision to make given the information that you had? It could be that if I knew the other person had three-of-a-kind, I wouldn’t have done that, but it was reasonable for me to think that the probability they had three-of-a-kind was only 10%. Therefore, I would make the same decision again despite it not working out for me in this one case.

KD: What’s the significance of the quantitative research projects for the assistant traders?

TS: The assistant traders do exactly what they’d do if they were actually trading. It’s different from mock trading, which is theoretical — they look at real market data and test their hypotheses to find ways to improve our trading. Several projects have been turned into strategies we’ve implemented or tools we’ve rolled out firm-wide, which is incredibly helpful.

However, the more important thing coming out of this is their quantitative research projects failing most of the time — they don’t uncover new ways to make money, which is a really good lesson. It’s not easy — there isn’t a lot of low-hanging fruit. Nobody can just say do this and have that lead to a money-making opportunity. Most of these ideas look and sound great, and when they start working on them, they may uncover a slight signal. But once you account for frictions in the market like cost of execution, slippage in the stock if you were to go trade it, or fees for using all the capital required to do the trade, then the opportunity goes away.

That’s a nice lesson on the difference between what happens theoretically in the classroom with figuring out a strategy that might work and what happens in the real world when you try to implement it.

KD: Moving away from SIG’s education, what have you been up to at SIG in the interim?

TS: There are a few things I trade now. I do block trades on some equity options through a specific relationship that we have. I know exactly why this risk is being transferred to us. I use a lot of the same pricing tools that we normally use, but with a large block trade, I’m not concerned with selection bias as I would be normally. That’s one thing I’m doing, but it takes up only a little bit of my time.

I’m also managing some of our private equity investments. I’m overseeing some of the relationships we have with strategic private equity investments — ones that read onto another part of our business in some way or other. 

The bulk of my time is spent in our insurance business. We have a Bermuda-based Class 3A reinsurance company and a US-based insurance brokerage group, and we’re passing the risk to the Bermuda-based reinsurance company. Most of the risks we’re looking at are shorter-term, chunky, more situational. We look at contractual bonuses in the sporting and entertainment worlds, contract frustration, event cancellation, price indemnification for lotteries, game shows and other mathematically-driven risks.

KD: SIG uses a lot of data, both external and internal, and stores it in multiple clusters using GPFS. Tell me about that.

TS: It’s important to ensure we’re not put out of business because of a single error. There’s a lot of redundancy in our storage. We care about power supply to our data center and ensure we understand the risks of everything from our battery backup to having multiple lines coming in from different sources so that all our systems continue to run. There are lots of real world physical operational aspects that go into protecting this, in addition to ensuring we don’t have the ability to make errant trades on the back of either bad code or small errors magnified by our robust data structures.

KD: How does SIG deal with colocation and speed?

TS: We have lots of systems that are colocated. We have some intelligence on chips that execute trades at colocated sites before they’re bounced so that we don’t have to deal with the time it takes for the speed of light to go from New York to Philadelphia and back.

The entire effort has been driven by a strong technology team that made sure they understood the requirements and then found the appropriate hardware and built the right systems.

That said, there are parts of the game that aren’t as speed-dependent as others. There are some arbitrage relationships that move literally at the speed of light, but then there are longer-term positioning approaches that don’t require hyperspeed.

KD: To prepare for this interview, I watched a YouTube video of a Zoom talk you and Kristina Alimard gave in 2021 for the University of Virginia on cryptocurrency. How does trading cryptocurrency compare with trading more traditional asset classes? 

TS: There are a lot of similarities. During the pandemic, equity option markets ended up looking more like the crypto markets than the other way around. You’d see large numbers of small-dollar value trades expressing an opinion in one direction or another that would move the market. In the equity option markets, we saw this with GameStop, and we saw similar trading patterns in the crypto markets. Understanding the fundamental value proposition of decentralized finance and the accompanying coins is important, but then understanding human behavior becomes even more important when you’re looking at the mass behavior witnessed in the crypto markets. ■