Podcast: Goldman Sachs Exchanges: Great Investors
Raj Mahajan, global head Systematic Client Franchise interviews Renaissance Technologies CEO Peter Brown on July 27, 2023
I grabbed some excerpts from the transcript for future reference.
I include my own commentary here and there.
Newsflash: Money is a big factor in what people choose to do
Raj Mahajan: Seems that you were right at the vanguard of the machine learning movement in 1993. So, why did you leave an exciting career at IBM for a small financial company in Long Island that no one had ever
Brown: …Three things happened. First, Bob had a second daughter accepted to Stanford. But he couldn’t afford to pay for her to go to Stanford on his IBM salary. So, she had to go to the agricultural school at Cornell, which offered scholarships to New York State residents. The second thing that happened is we had a daughter born. And a third thing was that Jim then offered to double my compensation. After that offer, I came home. I took one look at our newborn daughter and realized I had no choice in the matter. So, the decision to leave computational linguistics for a small hedge fund that no one had ever heard of was made purely for financial reasons.
Examples of Emotional IQ
[Kris: The EQ vs IQ thing is a false dichotomy. I suspect they are actually positively related but when we look at outliers on either dimension there is a major Berksons Paradox effect. RenTec has the reputation of being the true “smartest guys in the room” in the IQ/STEM sense of the word. And yet, multiple times in this interview I am struck at how people-savvy they have been. Which makes perfect sense to me. In a domain where the competition constantly learns and psychology plays an enormous role this is exactly what you expect. Only the naive who believe that investing is physics as opposed to biology cling to Spock-like caricatures of effective quants. Here are several excerpts demonstrating an deep understanding of human behavior]
Selling an approach to employees
Brown: At the end of 2002, Bob and I also took over the rest of the technical side of the firm, which included the trading of currencies, bonds, options, and futures. Now, our plan was to use the equities code that we and others had developed to trade these other instruments. But we recognized they would not be so great for morale to tell, say, one of the futures researchers, “You know all that code you spent the last decade of your life developing, guess what, we’re going to throw it out.” So, we had to spend quite a bit of time getting everyone to buy into our plan. To do this we used an approach that I learned from a biography I’d recently read of Abe Lincoln, which was to get them to come up with our plan themselves. Now, that took some time, but eventually it all worked out.
Jim Simons weighing the input to manage a risk crisis
See below: 2007 — “Quant Quake”
Jim Simons reading a situation shrewdly
Brown: In the fall of 2008, the whole financial system was stressed. So, we were concerned with the stability of our counterparties. So, we spent a lot of time with those counterparties and examined their CDS rates and so forth. I remember at one point, two senior executives from some firm we did business with came into our New York City office to meet with us. They assured us that the funds we had in our margin account were safe with them. And I was inclined to believe them. Why not? But after the meeting, Jim said, “Peter, they wouldn’t have come to our office. They wouldn’t have requested the meeting unless they were in real trouble. It’s time to get out.” So, we did. And Jim was right because shortly thereafter, that firm just disappeared.
Examples of automation and innovation within RenTec
Brown: When we got control of the New York office, the first thing I did was to walk around that office, find out what everyone was doing. And what I found was that many people were doing jobs that could be automated. So, we set out on a massive campaign to automate our back-office operations. We moved from checks and wires to SWIFT and ACH. We replicated counterparties margin calculations. We built a large legal database that could be accessed by computers to fill out regulatory forms. We brought in AI systems to automatically read and pay invoices. We automated the treasury department so that cash and margin needs could be managed by computers instead of humans. My point of view was that Stony Brook produces a huge list of transactions and New York City produces monthly statements, K1s, and government filings. And I just didn’t see why humans need to be involved in the process of translating trades to monthly statements. Now, 13 years later, we’re not done yet. And I’m embarrassed to admit that we still even have a few people who use Excel. But we’re getting there. In fact, I was told recently that we’ve eliminated 97 percent of the spreadsheets that had originally been used in the company.
Stories about risk management
March 2000 — Dot Com
Brown: Let me start with March of 2000 when the dotcom bubble burst. We were doing extremely well back then. And we had large positions in the internet stocks. They were traded on NASDAQ. At one point the head of risk control came to me and said he was worried about the size of our NASDAQ positions. But I told him not to worry, the computer knew what it was doing. Then we took a big loss one day. So, I worked through the night trying to understand what was going on. The next day we took another big loss. And I, again, worked through that night. So, now it’s the third day and I hadn’t slept for, I don’t know, 48 or 50 hours. And I was sitting in a meeting with Jim and a few others when the head of production knocked on the door and asked to speak with me. I walked out of the meeting, and he told me we were down again by a large amount. So, I walked back in the meeting, and I must have turned white or something because Jim took one look at me and said, “It doesn’t look good.” Now, not having slept the previous two nights, I remember thinking I’m not sure I can get through this. But I really didn’t have much choice in the matter. And so, we got back to work and eventually we did get through it. A couple days later I went into Jim’s office and told him that I’d screwed up in not appreciating the risk we were taking and said that if he wanted me to resign, I would resign. But he responded, “Peter, quite the opposite. Now that you’ve been through such a stressful losing period, you’re far more valuable to me and to the firm than you were before.” Now, that response really tells you something about Jim Simons.
2007 — “Quant Quake”
Brown: When that happened, I was on vacation, and I was on a very long flight back to Newark Airport. And the moment the plane landed, my phone went nuts with all kinds of texts and missed phone calls. So, I called into work when it was going on and I got Kim, Jim’s assistant. And she said, “Jim wants you to get back here as soon as you’re physically able.” So, I raced out. I found a taxi, leaving my family to fend for themselves at Newark Airport. And pushed the driver to drive as fast as he could from Newark to Long Island. I ran into my office, and I found Jim, Bob, Paul Broder, who was head of risk control, all holed up. And the office was full of cigarette smoke. I could barely breathe. And then there was this, I remember seeing this, 16 oz cup full of Jim’s cigarette butts. And I’m thinking, like, why do they have to do this in my office? And they were all staring through the haze at the computer screens trying to figure out what was going on. And Jim was interpreting every little wiggle and various graphs. He was really scared. And he wanted to cut back and hard. Paul also wanted to cut back. Raj, I’m sure you know, the head of risk control always wants to cut back. Because he doesn’t get paid to make money. He gets paid to make sure you don’t lose money.
And Bob, you know, Bob’s always very calm. But he wasn’t against cutting back. But I looked at the data and saw that the model had these enormous predictions, the likes of which I had never seen before. It was clear to me what was going on. People were dumping positions that were correlated with their own positions. And they were driving prices to ridiculous levels. I felt they had to come back. I argued that we should not cut back. That this was going to be the greatest moneymaking opportunity we’d ever seen. And if anything, we should increase our positions. But it was three against one. And so, we continued cutting back. But I succeeded somewhat because we cut back at a slower pace. And then at one point, miraculously, the whole thing came roaring back. And indeed, it was an incredible money-making opportunity. Now, what we learned from that was to always make sure we have enough on reserve to just hang on. Later, when Jim was about to retire, I reminded him of this period and asked if he was concerned that I was going to be so aggressive that I was going to blow the place up. But Jim responded that the only reason I was so aggressive was because I knew he was determined to reduce risk, another example of Jim’s insight into human nature.
What RenTec does differently
Brown: I guess there are some firms that make it their business to learn how others make money and try to learn their secrets. That’s not our style. We just hire mathematicians, physicists, computer scientists with no background in finance and no connections with Wall Street.
A few principles we follow:
The company was founded by scientists. It’s owned by scientists. It’s run by scientists. We employ scientists. Guess what, we take a scientific approach to investing and treat the entire problem as a giant problem in mathematics.
[Kris: In chatting with a friend who has proximity to RenTec, I learned of this a few years ago. I was intrigued by how they felt quite comfortable incubating highly promising individuals by offering a well-paying collegiate atmosphere that offered an alternative to traditional academia. It feels like just another instance of what I call risk absorption. RenTec is a highly efficient “bidder” for the risk of a scientist’s effort panning out. They can build a portfolio of talent in the form of a skunkworks knowing that they can scale important discoveries across their trading. Not unlike how a military R&D department might think of investments in scientists.
Science is best done through collaboration. If you go to a physics department, it would be absurd to imagine that the scientist in one office doesn’t speak to the scientist in the office next door about what he or she is working on. So, we strongly encourage collaboration between our scientists. For example, we encourage people to work in teams. We constantly change those teams up so that people get to know others within the firm. We pay everyone from the same pot instead of paying different groups in accordance with how much money they’ve made for us and so forth.
We want our scientists to be as productive as possible. And that means providing them with the best infrastructure money can buy. I remember when I was at IBM, there was this attitude that programmers were like plumbers. If you need a big project done, just get more programmers. But I knew that some programmers were, like, ten times or more productive than others. I kept pushing IBM management to recognize this fact. But it did not. I remember being in an IBM managers meeting and some guy from corporate headquarters was explaining how they created something called their headlights program. The goal of which was to identify the best programmers in the company and pay them 20 percent more than the other programmers. Now, I figured this guy from corporate was making, like, $300,000 a year. So, I raised my hand and suggested they increase the pay of their best programmers to $400,000 a year. And he was stunned. He said, “What? More than me? You’ve got to be kidding me. Well, if the guy’s Bill Gates.” I said, “No, Bill Gates was making, like, 400 million per year. Not 400,000.” Anyway, they just didn’t get it. We don’t make that mistake. We pay our programmers a ton in accordance with the value we place on the infrastructure they produce.
- No interferenceWe don’t impose our own judgment on how the markets behave. Now, there’s a danger that comes along with success. To avoid this, we try to remember that we know how to build large mathematical models and that’s all we know. We don’t know any economics. We don’t have any insights in the markets. We just don’t interfere with our trading systems. Yes, of course there are a few occasions where something’s going on in the world and so we’ll cut back because we think the model doesn’t appropriately appreciate the risk of what’s going on. But those occasions are pretty rare.
- Time We’ve been doing this for a very long time. For me, this is my 30th year with the firm. And Jim and others were doing it for a decade before I arrived. This is really important because the markets are complicated and there are a lot of details one has to get straight in order to trade profitably. If you don’t get those details straight, the transaction costs will just eat you alive. So, time and experience really matters.
A word on politics
[Kris: Peter Brown is liberal and co-CEO Bob Mercer is famously conservative. I can say that coming from the trading world, the liberal perspectives are in the minority amongst the traders but less so amongst the academics.]
Raj Mahajan: Is it true that while Bob Mercer and you have different politics, you worked closely for nearly 40 years at IBM and Renaissance?
Peter Brown: Yes. It’s true. Bob and I began working together at IBM 40 years ago. And for most of the time, we’ve had offices right next to one another. So, we’ve done a lot together. And we’re still really close. In general, I find no better way of building friendships than through the collective creative process of building something together. And I see no reason why politics should interfere with friendship.
Man vs machine stories
1) My understanding is that you had nothing to do with finance until age 38 and, instead, began your career working on automatic speech recognition. How did that happen?
Brown: So, at one point during high school I learned about the Fast Fourier transform. And I thought this was about the coolest thing I had ever seen. Probably because I went to an all-boys’ school and had nothing better to contemplate. Anyway, for some reason I got into my head that with the Fast Fourier transform it should be possible to recognize speech. You just take the speech data, transform it into the frequency domain. Match it up against patterns for words. And presto, magic, HAL would be born. And this idea always stuck around in the back of my mind.
Then when I went to college I majored in math and physics. But in my senior year I had to fulfill a distribution requirement. So, I took a course in linguistics. And one day in the back of that course I heard a couple students talking about some guy whose name was Steve Mosher who started a company called Dialogue Systems that was doing speech recognition. And I thought, wow, great, I remembered this idea from back in high school. After class I raced over to the physics library. That’s because this was before the internet, so you had to go to the library. And I looked this guy up. And I found a paper he’d written. And I tracked him down. Applied for a job. And he hired me. And when I was there, I just fell in love with the idea that through mathematics it might be possible to build machines that do what humans do. I just loved the idea of exposing human intelligence to be nothing more than robotic computation.
2) I recently heard that in a talk you give at Harvard Business School you mentioned that you had a role in starting up the Deep Blue project at IBM. Can you tell us about that?
Brown: Wow. Okay. I had been at IBM for a year or two. And I was standing in the men’s room one day when the vice president of computer science, a man named Abe Peled walked up next to me. I thought to myself, now’s my chance. I turned to him and said, “Dr. Peled, do you realize that for a million dollars we could build a chess machine that would defeat the world champion? Think of the advertising value to IBM.” He turned to me, looking kind of annoyed, and said, “What’s your name?” So, I told him. And then he said, “Could you please let me finish up here?” And so, I thought, wow, I had made a big mistake. So, I apologized, and I high tailed it out of there as fast as I could hoping he’d forget my name even faster.
But a half hour later, he called me in my office and told me that if I wanted to build a chess machine, he’d put up the million dollars. I told him that I was occupied with speech recognition. I have three friends from graduate school who could build it. He said, “Okay, hire them.” So, we did. They built the machine. I named it Deep Blue. In the first match, the IBM machine was a very weak machine. Weak physically. You know, I think only one special purpose chip in it. And we lost. The final match, however, was a different story. IBM had a much, much stronger machine with hundreds of special purpose chess chips. IBM won that match and IBM’s stock jumped $2 billion afterwards. Of course, it fell back down later.
Now, a few years ago I was asked to speak at the Harvard Business School. And when I arrived, outside the auditorium, I could see all these protesters. And I thought, oh no, why are they protesting me? What have we done? Is there something I’m not aware of? I really didn’t want to do that. But as I got closer, I could see they were all holding signs about investing in Puerto Rico. And I thought, what is this all about? I was totally confused because I didn’t think we had anything to do with Puerto Rico. Then it turned out that the speaker before me was some guy named Seth Klarman from some firm named Baupost. Evidently, that firm had some investments in Puerto Rico and the protesters were protesting him. So, I went in to see Klarman’s talk, or at least the end of Klarman’s talk, to find out what all the hullabaloo was about.
At the end of his talk, someone asked him his thoughts on quantitative investing. I suppose it was a set up for my talk. I don’t know. And I carefully noted his answer which was, “To do what I do takes a certain amount of creativity and finesse that a computer will never have.” And all those Harvard Business School MBAs seemed to really like that response. So, when it was my time to speak, right after him, I began by pointing out that after defeating Deep Blue in the first match, Kasparov was elated and gave a press conference at which he said, “To play chess at my level takes a certain amount of creativity and finesse that a computer will never have.”I then went on to point out that two years later we crushed him. Now, I’m not sure that’s how things will evolve. But whether it’s speech recognition, machine translation, or building large language models, or chess, or making investment decisions, I continue to love the process of showing that human intelligence, intuition, creativity, and finesse are nothing more than computation.
[Kris: In defense of Klarman, like the pod shops, I don’t think RenTec is investing so much as trading. Marc Rubinstein writes:
Dmitry Balyasny, founder of Balyasny Asset Management, attributes the model to a trading view of markets as distinct from an investing view.
“[Its] origins go back to my origins as a trader and thinking about how to build out business around trading… It makes sense to have lots of different types of risk-takers, because you have less correlation, you could attack different areas, the markets, and have specialists in different areas.”
Addressing Brown’s obsession with “exposing human intelligence to be nothing more than robotic computation.”
In The Introspection of Illusions, author David McRaney parses opacity of the intelligence and preferences buried in our subconscious:
Psychologically speaking, users found it easy to access the feelings that prompted them to give those films one star or five. Explaining why they made one feel that way would require the kind of guided metacognition that the Netflix interface simply couldn’t offer. Even when you stepped away from the code and the spreadsheets and asked people in person, they might not be able to tell you. They could make a guess. They could attempt to explain, justify, and rationalize their feelings, reactions, and star ratings, but without a conversational tool, a back and forth to get past all that to something honest and perhaps previously unexplored, you ran the risk of precipitating a psychological phenomenon known as the introspection illusion which would likely result in yet another phenomenon known as confabulation. There’s an entire literature of books and papers and lectures and courses devoted to this side of psychology. To put it very simply, we are unaware of how unaware we are, which makes us unreliable narrators in the stories of ourselves. You are, however, amazing at constructing stories as if you did know the antecedents of those things when explaining yourself to yourself and/or others.
There are parts of us we can’t access, sources of our emotional states we can’t divine, and I find some strange poetry in the fact that, like us, the algorithms can’t always articulate the why of what we do and do not like. Yet, through millions of A/B tests slowly zeroing in on more and more successful correlations, the Netflix Recommendation Engine can produce a glimpse of something a bit like the sort of profound, soul-exposing knowledge earned via an intense introspection that we could never achieve. Something a few fathoms deeper than “I don’t know, it just wasn’t for me.”
1) Is it true that at one point you went to IBM to suggest that the statistical methods you were using in speech recognition could be applied to finance, and asked to be given an opportunity to manage some fraction of IBM’s corporate cash?
Brown: Yes. I think that was in 1993. But IBM corporate had absolutely no interest. So, instead we went to Renaissance where we did the same thing we had in mind for IBM, but instead with money Jim Simons had raised.
2) Is it true that since you first joined Renaissance you have spent nearly 2,000 nights sleeping in your office?
Brown: Yes. My wife works in Washington DC. And my experience has been that when a husband and a wife work in two different towns, the husband commutes. Psychologically, if I’m going to be away from my family, I have to work. I sleep in my office when I’m in Long Island.
For me, productivity-wise it’s really fantastic being able to spend nearly 80 straight hours each week with no interruptions except sleep thinking about work before spending three more normal days at home. Of course, I really miss my family. But the freedom to concentrate nonstop on work while surrounded by my colleagues is hugely valuable. And the job is so demanding, I really don’t see how I could do it otherwise.
[Adds this] I’m just one of those types who can’t sleep. Not by choice. I just can’t sleep. So, I often am on the computer by around 2 am. And it’s true, I tend to send a lot of emails out in the middle of the night.
3) Is it true that you almost exclusively hire people with zero background and finance?
Brown: Yes. We find it much easier to teach mathematicians about the markets than it is to teach mathematics and programming to people who know about the markets. Also, everything we do we figure out for ourselves. And I really like it that way. So, unlike some of our competitors, we try to avoid hiring people who have been at other financial firms.
[Kris: The prop trading firms think similarly. My friend Joel talks about how Brown’s claim that it “is is easier to teach markets to mathematicians than it is to teach math to market experts, may seem dismissive to market-centric people but in reality is more of a statement about what “math” is at Renaissance.” He goes on to distinguish about levels of math but I latched on to this a more general observation:
Markets person isn’t a thing. Markets thinking is systems thinking and anyone from any discipline can learn that. From there go on Investopedia and learn how a zero coupon bond or share of stock works. start with a good, teachable mind then label the variables.
Math/STEM skills are legible markers of computational/rigorous thinking. Someone trained in the nitty gritty of assumptions, what follows, and so on. Making abstractions concrete.
If I’m generous it took a month of professional training for non-finance STEM grads at SIG to know everything finance grads would have brought to the table. But you can’t teach math and computer science in a month.
Ultimately this is only part of the story of getting a great start in finance. There’s a Berksons Paradox once you are in the pool of high level finance employment where the math skills don’t correlate as much with talent. You get older and realize the dichotomy of being a math person vs a verbal person that you carried as an identity when you were young is bullshit. Skills in either are likely highly correlated. But maybe the right door or guidance wasn’t there to help you see that.]
4) What do you actually look for in applicants?
Brown: Math ability. Programming ability. A love for data. A work ethic. And most importantly, the ability and desire to work will in a collegial environment.
5) How do you actually assess those qualities?
Brown: I think probably the same way other firms do. First, we get resumes. Those that look promising we give them phone interviews and we ask them for references. If those pan out, then we invite the promising applicants to give research talks. Talks like if you’re applying for a job at a university or something like that. And then we put them through a grueling day of solving problems in math, physics, statistics, computer science, and so forth at a blackboard.
6) Is it also true that your staff had to install mirrors in the corners of the office to prevent you from flying into people as you rode a unicycle around the office?
Brown: Where did you get all these questions from? Yes, it’s true. Although, I don’t ride a unicycle anymore because at one point I crashed and the unicycle broke.