ReSolve Riffs with Michael Robbins: Don’t Trust a Quant Unless…

In this conversation, we delve into the world of quantitative investing, exploring its intricacies, misconceptions, and the common mistakes people make when trying to apply it. The discussion also sheds light on the importance of asking the right questions and setting up the problem correctly in quant investing.

Topics Discussed

  • The common misconception that a single tool or program can solve all problems in quant investing
  • The importance of setting up the problem correctly and asking the right questions in quant investing
  • The challenges of communicating complex quant concepts to laypeople or investment professionals without a background in quantitative methods
  • The role of intuition and experience in quant investing and how they still hold significant value over calculations
  • The use of quantitative methods in large institutions and the need for human supervision to prevent unusual or excessively risk-taking actions
  • The impact of back testing and the importance of incorporating experience into it
  • The discussion on whether machines can completely take over quant investing and the current limitations
  • The exploration of arbitrage opportunities in quant investing and how they can be exploited by mid-sized banks

This conversation is a deep dive into the world of quant investing, offering valuable insights into its complexities, the common pitfalls, and the role of human intuition and experience. It’s a must-listen for anyone interested in quantitative investing, providing strategies to navigate this complex field.

This is “ReSolve Riffs” – live on YouTube every Friday afternoon to debate the most relevant investment topics of the day, hosted by Adam Butler, Mike Philbrick and Rodrigo Gordillo of ReSolve Global* and Richard Laterman of ReSolve Asset Management.

Listen on

Apple Podcasts

Listen on


Subscribe On



In the realm of quantitative investing, misunderstandings abound. Many believe that a single tool or program holds the secret to success, but a well-defined problem and appropriate question are the critical first steps. Skipping this vital phase can lead to a ‘cliff’ scenario where substantial effort must be retrospectively corrected. Even with the aid of quantitative analysis, human supervision is necessary to prevent excessive risk or unseen anomalies. As seen in fields such as pathology, even professionals can err when solely relying on AI. It’s clear then that humans must remain involved in decision-making to help avoid irretrievable mistakes. A significant challenge faced in finance is the understanding and application of quantitative techniques. Currently, resources fall into two camps: academic texts that are inaccessible to non-quants, and oversimplified books providing minimal practical guidance. This leaves a knowledge gap for those wanting to apply quantitative strategies. There’s a need for resources which effectively bridge this gap, providing practical guidance understandable to both C-suite executives and decision-makers, who often lack a background in quantitative methods. The practical application of quantitative techniques to finance is fraught with misunderstanding, especially outside specialized hedge funds and big banks. The challenge lies not only in the complexity of finance data but also in the transition from a technical role to a portfolio management or CIO role. There is a need for resources that can bridge this gap, providing guidance and insights that are understandable and usable by non-technical decision-makers. Communicating quantitative concepts to non-quants is often challenging. While machines can crunch numbers with incredible speed, the intuition and experience of senior executives are still highly valued in decision-making. They can quickly identify if an answer is wrong without the need for calculations. While there are areas where technology is essential, such as high-frequency trading or NLP, decision-making in larger organizations still requires human intuition and validation to prevent excessive risks. Back-testing is a crucial tool but must be used considering real-world constraints like market impact and concentration risk. Understanding these factors can lead to more effective strategies that align better with market realities. However, communicating these complex concepts to non-technical individuals remains a challenge, highlighting the need for resources that bridge this gap between technical and non-technical decision-makers. In conclusion, while quantitative methods can lead to better answers, the role of human supervision and experience is still vital in ensuring sound decision-making. The combination of technical and non-technical inputs leads to better outcomes, as seen in the pathology example. There is a pressing need for resources that bridge the gap between technical knowledge and practical implementation, making quantitative techniques accessible and usable by decision-makers. Furthermore, technical and non-technical decision-makers need to work towards better communication and understanding so that quantitative techniques can be applied effectively in finance.

Topic Summaries

1. Misconceptions about quant investing

Misconceptions about quant investing often lead people to believe that there is a magic program or tool that can solve all their problems. However, the reality is that setting up the problem and asking the right question is more important than finding the solution. This misconception can have damaging consequences, as going back and redoing everything to solve a different problem is technically called a ‘cliff’’. Many people don’t even consider the question until they are too far into the solution, making it difficult to rectify any mistakes. Quantitative methods may lead to better answers most of the time, but human supervision is necessary to ensure that nothing unusual or excessively risky is happening behind the scenes. While it would be ideal to have exceptionally self-disciplined individuals at the helm, studies have shown that even professionals like pathologists can make mistakes when relying solely on AI programs. Therefore, it is crucial to have humans involved in the decision-making process to avoid irrecoverable errors. Overall, the misconception that a single tool or program can solve all problems in quant investing can lead to costly mistakes and the need to start over, highlighting the importance of careful problem setup and human supervision.

2. Lack of understanding in applying quantitative techniques to finance

In the world of finance, there is a significant lack of understanding when it comes to applying quantitative techniques. While there are numerous books available on the technical aspects of quant investing, they often fall into two extremes. On one hand, there are academic texts that delve into complex mathematical concepts, making them inaccessible to non-quants and decision-makers. On the other hand, there are oversimplified books that fail to provide practical guidance on how to apply these techniques to finance. This leaves a gap in knowledge for practitioners who are looking to leverage quantitative strategies in their decision-making processes. The participants acknowledge that there is a need for a book that bridges this gap and provides accessible information for C-suite executives and other major decision-makers. They emphasize the importance of writing a book that they themselves would have wanted to read 20 years ago, one that explains how to apply quantitative techniques in a way that is understandable and applicable to the finance industry. The participants also discuss the challenges of communicating quant concepts to individuals without a background in quantitative methods. They note that during the COVID-19 pandemic, they observed a lack of understanding and receptiveness to complex problems, making it difficult to effectively communicate these concepts. Overall, the discussion highlights the need for better understanding and application of quantitative techniques in finance, and the importance of bridging the gap between technical knowledge and practical implementation.

3. Challenges in communicating quant concepts

Communicating quantitative concepts to non-quants or decision-makers who are not proficient in quantitative methods can be challenging. The language and intuition gained from experience in quantitative finance are often difficult to convey, and there is a need for better communication and understanding between technical experts and decision-makers. Many people, especially engineers, mistakenly believe that there is a magic program or tool that can solve any problem. However, setting up the problem correctly and asking the right question is more important than finding the solution. This delicate process is often overlooked, leading to wasted time and effort. Additionally, there is a lack of understanding outside of specialized hedge funds and big banks on how to apply quantitative techniques to finance. Finance is a complex data set that often violates the assumptions of traditional models, requiring expertise in working with its idiosyncrasies. Transitioning from a technical background to portfolio management or a CIO role can be challenging due to the gaps in education and knowledge. The book discussed in the conversation aims to bridge this gap and provide insights for non-technical decision-makers. However, communicating quantitative concepts to laypeople can be difficult, especially when they have preconceived notions or are defensive. Trusting intuition and relying on human validation is still prevalent in decision-making, particularly in smaller firms. Overall, there is a need for better communication and understanding between technical experts and decision-makers in order to effectively apply quantitative approaches in finance.

4. The role of intuition and experience in decision-making

Intuition and experience continue to play a crucial role in decision-making, even in the field of quantitative finance. While quantitative methods can provide more accurate answers in most cases, human supervision is still necessary to prevent unusual or excessively risky decisions. The discussion highlights that intuition and experience are highly valued in decision-making, especially in smaller organizations where trust in human judgment is still prevalent. The participants emphasize that although technology and data science have made significant advancements, they have not yet reached a stage where they can outthink humans. The conversation suggests that while machines can take advantage of scale, complexity, and speed, they are not ready to overtake the world. The senior people in organizations still have a significant influence on decision-making, relying on their intuition and experience. However, the participants acknowledge that there are certain areas, such as NLP, data scraping, and high-frequency trading, where reliance on technology is inevitable due to their technical nature. Nonetheless, decision-making in smaller firms often involves validating decisions through human judgment. The discussion also highlights the importance of experience in decision-making, drawing parallels with the field of pathology, where a study found that a combination of human pathologists and AI programs yielded the most accurate results. In conclusion, while quantitative methods offer better answers most of the time, human supervision and experience are still essential to ensure sound decision-making in finance.

5. The limitations of quantitative methods

Quantitative methods have their limitations, and there are still areas where human intuition and experience are more valuable. Machines can take advantage of scale, complexity, and speed, but for big decisions and complex situations, human input is still crucial. The singularity where computers can outthink people is yet to come. People often have misconceptions about quantitative methods, thinking that there is a magic program that can solve any problem. However, setting up the problem and asking the right question is more important than the actual solution. Many people don’t realize this until they are too far into the solution. While machines can calculate and provide answers quickly, intuition and experience are still highly valued in decision-making. Senior people in organizations still have a significant role in decision-making, as they have the ability to quickly identify if an answer is wrong without needing to calculate it. While there are opportunities for machines to take advantage of scale and complexity, they are not yet ready to take over the world. In larger organizations, there is a reliance on technology for technical tasks like NLP or high-frequency trading, but decision-making still relies on human intuition and validation. This is because humans have the ability to avoid irrecoverable mistakes and make judgment calls in complex situations. Quantitative methods may lead to better answers most of the time, but human supervision is necessary to ensure unusual or excessively risky decisions are not being made.

6. The importance of considering real-world constraints in backtesting

When conducting backtesting, it is crucial to consider real-world constraints and factors that may impact the execution of a strategy. Backtesting is a valuable tool, but it can produce unrealistic results if these factors are not taken into account. One important factor to consider is market impact, which refers to the effect of large trades on the market. It is difficult and expensive to directly measure market impact, but it can be inferred through the use of microstructure models. By understanding the impact of large trades, one can spread out the execution of trades over time to minimize the risk of the market drifting away. Another factor to consider is concentration risk, which arises when a portfolio is heavily weighted towards a few assets. This can be a problem, as the performance of these assets may be highly correlated, leading to increased risk. Experienced individuals are more likely to build these factors into their backtests, as they have a deeper understanding of the market and the nuances involved. They are able to connect the dots and make informed decisions based on their experience and intuition. Backtesting should not be seen as a standalone tool, but rather as a part of a larger process that takes into account real-world constraints and factors. By considering these factors, one can design more robust backtests that are better aligned with the realities of the market.

7. Challenges of communicating quantitative concepts to laypeople

Communicating quantitative concepts to laypeople or individuals without a background in quantitative methods can be challenging. Laypeople may struggle to grasp complex concepts, and there is a need for better language and communication strategies to bridge the gap between technical experts and non-technical decision-makers. The conversation highlights the intention to write a book that is accessible to non-quants and higher-level executives, aiming to provide the information that the participants wished they had known 20 years ago. While there are many excellent books on the technical aspects of quantitative finance, there is a lack of understanding outside of esoteric hedge funds and big banks on how to apply these techniques to finance. The participants also discuss the types of quant strategies that are most appealing to larger organizations, with a reliance on technology for technical tasks like NLP, data scraping, and high-frequency trading. However, decision-making at smaller firms still relies on human intuition and validation. The example of car accidents is used to illustrate the importance of human decision-making, as people still trust their intuition despite the availability of quantitative models. The conversation also touches on the challenges of communicating quant concepts to laypeople. The group admits to not being experts in teaching these concepts and expresses a loss of faith in peoples’ ability to understand complex problems, particularly evident during the COVID-19 pandemic. Overall, the discussion highlights the need for better communication strategies to effectively convey quantitative concepts to non-technical individuals.

8. The role of experience in designing robust quantitative models

Experience plays a crucial role in designing robust quantitative models. Understanding the subtleties of the market, considering known unknowns, and accounting for real-world constraints are essential for creating successful models. Experience helps in avoiding common pitfalls and designing strategies that can withstand real-world challenges. Quantitative methods, such as NLP, data scraping, and high-frequency trading rely heavily on technology and technocrats. However, when it comes to decision-making in smaller firms, intuition and human validation still hold value. While quantitative methods provide better answers most of the time, human supervision is necessary to prevent unusual or excessively risk-taking behavior. This is because making a mistake in certain contexts can have irrecoverable consequences. Backtesting is an example where experience plays a significant role, as it requires understanding the nuances of the data and methods. Experience allows for the development of intuitions and insights that are difficult to articulate but can lead to successful strategies. Overall, experience complements quantitative methods by providing a supervisory role and ensuring that models are robust and aligned with real-world dynamics.

Michael Robbins, CFA

Michael Robbins is a professor at Columbia University where he teaches quantitative investing including graduate classes in Global Macroeconomic Tactical Asset Allocation (GTAA) and Environmental, Social, and Governance (ESG) Investing.

He is a leading member of the Board of Directors of Blythestone Corporation and on the Board of Directors of Urban Edge Capital hedge fund (UEC). As a quantitative expert, Robbins develops investment models that allow money management firms to invest in large numbers of stocks to produce positive returns.  He provides oversight and offers advice on governance, strategy, and investments.

Robbins started his career in nuclear physics at RTII and later worked in electronic warfare at a subsidiary of Northrop. He then became a proprietary trader at Oppenheimer/CIBC, with $350 billion of assets under management.  There he managed portfolios while generating substantial profits and overseeing no losing years. 

After working as a prime loans and derivatives portfolio manager in Washington Mutual’s treasury department, he moved to ECR Capital Management, a hedge fund. He later became the Chief Investment Risk Officer at Utah Retirement Systems (state pension systems), with $16 billion in assets under management. Returning to ECR as their Chief Investment Officer, he advised the senior management of National Australia Bank ($685 billion in assets) on asset management and risk analysis. He also advised large family offices, major banks, and other large financial institutions. He then joined Spruce Investments as their Chief Investment Officer, managing billions in funds. Under his leadership, Spruce was nominated for Best Global Equity Fund & Best Global Macro Fund. Bleakly Financial Group recruited him as their Chief Investment Officer, where he managed twice as much money as he did at Spruce. Later he became the Chief Investment Offer for FNB (FirstRand), a South African bank that is the largest in the continent with 8.5 million clients.


Quantitative Investing – a Definition

[00:00:00]Michael Philbrick: Michael, quant investing is not what most people think it is. How do you think most people view quant and what are the key differences to traditional approaches?

[00:00:11]Michael Robbins: Well, I think the biggest difference, well, the biggest way that people misperceive quant is that they think it’s about getting an answer, about getting a number. But if you ask anybody who went to school for physics or engineering, they’ll probably tell you that it’s just a thought process. The way of approaching a problem and methodically figuring out how to solve it is, I think, the biggest benefit to Quant in general.

And the number itself is almost not even important because things change and the relevance of the number might not matter after a few moments of market action. So I think it’s really the way we go about solving a problem and people who do it intuitively, they can be brilliant and come up with a better answer, but there’s not that much structure and comfort as there is in quant investing.

And that leads to kind of those misconceptions, follow through towards, like, what mistakes people make when they try to do it. Like, people generally think that, especially engineers, think that they can get one tool that’ll solve a problem. This is magic program, you know, just throw ChatGPT at it and everything will be fine.

And that’s very much not the case. Setting up the problem is very delicate and much more important than actually solving it. Specifically, asking the right question is a very difficult thing. And a lot of people don’t even think about the question until they’ve gotten so far into the solution that it’s very damaging to go back and redo everything to solve a different problem.

And that’s technically called a specification problem. And probably that’s the most common problem I see in the hundreds of tech projects that I’ve overseen over the past few years.

[00:02:12]Michael Philbrick: Fabulous. Well, that sets the table for a great conversation today. And I think just before we start, we’re going to do some housekeeping. One is this is not investment advice, rather for educational purposes. And we’re going to go down some rabbit holes and things like that. So, you know, for guys on YouTube, I’m not sure you should get investment advice there, but we’re going to have some fun times education, educating you, I hope. Today’s Riffs also is brought to you by ReSolve Asset Management. Please check out us at Also You can view our mandates that we run across various ETFs, mutual funds, and private pools in those locations. And what you just heard was, joining us today, Michael Robbins.

He is and has been a well traveled CIO who’s managed pensions, endowments, family offices, worked at major banks, is a professor at Columbia University teaching a myriad of graduate class courses related to the field of investing. So we have truly a galaxy brain here today. And we’re digging into his new book, Quantitative Asset Management. You can learn more about that at www.quantitativeassetmanagement. com. And we’re going to get into it. Cue the music.

It’s time to get out the shovel and the pick. Let’s get…

Quantitative Asset Management – The Book

[00:03:28]Rodrigo Gordillo: You snatched up that URL, eh? That’s a perfect URL for what you do. It’s amazing. I’m shocked. Transcripts provided…

[00:03:35]Michael Robbins: I was very happy to get that. I couldn’t believe nobody had it before.

[00:03:40]Adam Butler: You didn’t buy it like 10 years ago in anticipation of writing this book. I can’t believe it was, it was still available.

[00:03:45]Michael Robbins: Yeah. It was very last minute, but, uh, I got lucky.

[00:03:50]Adam Butler: This book. I can’t believe you got, you got a recommendation from none other than Frank Fabozzi. Anyone who’s gone through the CFA program will know that Frank wrote probably half the books, certainly all the books on quantitative methods, and I think most of them on fixed income as well. So, and I think he heads up, still heads up the CFA Research Institute if I’m not mistaken. If not, he did so for decades. And Mark Baumgartner.

[00:04:20]Michael Robbins: …like that

[00:04:22]Adam Butler: What’s that?

[00:04:23]Michael Robbins: The Journal of Data Science as well, I believe.

[00:04:26]Adam Butler: Yeah. Yeah. I mean, yeah, he’s omnipresent and, in the field of empirical finance. So that’s pretty neat. And Mark Baumgartner. And this is a 485, 490 page beast, right? So congratulations, a heck of an achievement. Why did you feel, um, that, you know, you were so motivated to write a book of this scope? Obviously you’re not, it’s not the first book on quantitative asset management. What is different about your approach and the subject matter and how you approach this book from some of the canonical, other canonical books in the field?

[00:05:05]Michael Robbins: Yeah. There, there’s a bunch of really interesting questions in what you just said. I’ll try to, uh, unpack them. First of all, I really want to thank the people that gave endorsements for the book. I was overwhelmed and humbled by the quality of the endorsements I got. I don’t deserve them. Every one of them, they’re just wonderful.

And the, and, and the size of the book, it was actually about three times the size before editing. And that’s kind of a problem with the book. Some of the criticism I get is that it’s not in depth enough. And the reason why it’s not in depth enough, it is the last answer to your question that I and my students and other people try to solve real quant problems.

They Google these problems and for various reasons, the different examples on the Internet and in books, are not really sufficient to solve a real problem. If you look up say, an optimization problem, they usually talk about a few stocks or ETFs. If you want to Google the Kelly criterion, it’s almost impossible to find an example for multiple assets, right?

Like they just aren’t real examples you can dig your teeth into and you kind of get a hint of how to solve it, but it doesn’t take you over the finish line. So I wanted to bring that to bear, which made the book really, really long because these are very involved answers. And then when we cut the book down, I had a choice to making it very narrow and deep, or making it a little wider.

And I chose to bring up a lot of examples that people should be thinking about, especially people who are learning about finance or also professionals who specialize, because most people on Wall Street have a specialty and they know everything there is to know about that narrow bit and they don’t even realize what they don’t know.

And that’s what happened to me. I started on Wall Street as an arbitrageur. And I became an expert at a very technical trade. And then later in my career, when I became a CIO and a CRO and did other things, I realized there’s just so much that I had no, I didn’t even know to ask the questions about. So I wanted to open people’s eyes.

Not necessarily give them everything they need to solve the problem, but at least give them all the questions they need in order to look for the answers. And that was really the motivation behind it. I didn’t see really a lot out there. Either they were very academic and almost impossible to apply for a practitioner, or they were just overly simplistic and it just didn’t really show what needed to be shown.

[00:07:58]Michael Philbrick: I think that’s a conversation that I think we would echo those comments where we’ve had many discussions with portfolio managers and CIOs about the concepts in your book and others. And lots of people lack that global understanding of the structural deficiencies that exist in portfolios, right?

Have you fully exploited diversification? Are you balancing the risk premiums that you’re harnessing? Are you allowing the market to dictate the risk and structure of your character of the portfolio, or are you taking some active role in making sure that, that you’re balanced, diversified, and managed on the risk side of it?

So I think that we would echo those set of sentiments loudly, and I think something else I want to get into a little later, but I’ll turn it back over to Adam is like, how do you get the investor, the end user, the board to buy into these factors that are so different and differentiated from everybody else.

But I’ll, I’ll side, I’ll side that for a moment and throw it back to you, Adam, because I think you’re on a bit of a roll there.

[00:09:03]Rodrigo Gordillo: That’s why he was, he was in a role. Now he’s fully frozen. You’re back, Adam.

[00:09:11]Adam Butler: I’m just that still, what do you mean that’s, I was just being that still, yeah, actually that’s a good segue, Mike, because I was going to ask, because I know having sort of gone through your background and having read through, especially focused on the initial kind of introduction sections of the book, it seems like you’re speaking to two decision makers at higher levels in the organization, that are typically the ones that are maximally proficient at the actual quantum, like technical quant tasks and thinking that are, you know, that typically go into the analysis and the development of investment strategies. Was that on purpose? Did you want to write a book that was accessible to non- quants and, you know, higher level C-suite executives and other major decision makers at these institutions, rather than just speaking to technical engineers.

[00:10:10]Michael Robbins: Yeah, I, the way I approached it, uh, and I hope it came across this way was I was writing a book for my former self, but what should I have, you know, wanted to know 20 years ago. And there are lots of excellent books on the technical aspects and people much smarter than me are writing them. But there’s a big lack of understanding outside of, you know, the really esoteric hedge funds and prop desks and big banks, of how to apply those techniques to finance.

Because finance is a really difficult data set. There’s a, as I’m sure you know, there, there’s a lot of things about it that violate the normal assumptions of most models. And so working with those idiosyncrasies is something that even really experienced data scientists don’t know because they are focused on the technicalities. Likewise. I think it’s, it’s a relatively common progression, to go from science to trading to portfolio management, and then eventually maybe being a CIO. And it’s really hard to get that education you need to fill in those gaps. I was really surprised at some of the things I learned over the years.

And I thought while I was doing that, I might also approach maybe a CIO who’s not a technical person, but manages them, and maybe give him a little understanding. Maybe he won’t read the whole book. Maybe you read the first third and then, you know, kind of get a background to help him manage these technical people.

So, I was just trying to fill that empty space between the high quality technical books and the plethora of trading books and trading psychology and all that stuff. Those already existed, but there wasn’t a lot to connect them.

[00:12:04]Adam Butler: This happens a lot, I think, at a lot of organizations where the people who rise into leadership roles are not necessarily those with the deepest technical skills. And so do you have any insights on how a manager that wasn’t sort of steeped in a quantitative background can be most effective at managing technical people and getting the most out of their potential? What are the pros of having a, being a manager who, you know, didn’t dedicate their life to engineering or quantitative methods, and then what are some of the disadvantages and how can, how could they be overcome, do you think?

[00:12:56]Michael Robbins: I heard a lot of anecdotes when I started out working on Wall Street and one of them was that, some people actually turned down the title of managing director when they’re promoted, because what happens is, your job changes. You’re supposed to be a technical expert, really good at whatever it is you were good at up until the point that you become an MD, and then your job is selling your product to management and to clients, right? And you’re no longer actually doing that stuff. You’re basically supporting the people who are doing it.

And a lot of technical people are not good at that. They’re not good at selling. They’re not good at communicating. And, the smart ones may realize that and say, I don’t want the promotion. I can do better focusing on my strengths. Other people take a shot and find out if they’re good communicators. One of my favorite bosses once told me that he felt it was his job to give me the resources I needed and to support me, and shelter me from all the political nonsense going on above his level, right? And there’s a lot to that. People, some of the viewers might not realize that within a large organization like a bank, it’s like there’s competing businesses, there are takeovers, there are people taking each other’s resources. If you are say a trading desk or a research arm, you literally have to sell your services to other departments.

And they do have the option to going and buying third party products, and it could be a whole political battle to get funding and to keep business and all that stuff. So, a lot of that, those things I didn’t really know about in the first half of my career. And as I moved into the C-suite, I realized, you know, how much adversarial, you know, infighting and, arguing and persuasion is involved. And I do make that point a bit in the book. A lot of people like to skip over the first third of the book where I describe how to create a business plan and investment policy statement and things like that. But I think they’re really great because they help you in that part of the job. They help you not only understand exactly what you’re doing and to do it properly, but also to explain it and defend it to other people who may not really have bought into it yet.

And especially when the market turns against you, and people want to pull the money out from underneath you, and you have to defend yourself and say, no, no, it’s, it doesn’t look like it’s working, but it’s really just fine. We were prepared for this. We tested for this, you know. We’ve got a plan and, you know, your worries are perfectly reasonable. It’s a human emotion, but we’re, we have a better vantage point because we’ve studied this long before it happened and you’re just reacting now that it’s occurring. Right. So I try to prepare people for that because it’s I think, a pretty common path to be a trader, to want to start a hedge fund or work for a hedge fund.

Maybe that fund only has a few people in it, a few, you know, management people, and then suddenly you’re catapulted up into that sphere and you have no preparation for that. You’re used to looking at a screen with a bunch of numbers and now you have to persuade people and you have clients who get upset when they lose money and maybe they even sue you for practically no reason at all, you know, and you have to deal with legal documents and, and all sorts of crazy things.

Uh, so I tried to incorporate that because books do talk about it, but I haven’t seen one book that went from beginning to end, right? From designing your product in your business, to building your models, to managing your models, and to actually running the business. I was just trying to connect all the points.

[00:17:03]Rodrigo Gordillo: I mean…

[00:17:04]Michael Philbrick: Yeah, I think the one thing that resonates with me on that side too, is some of the things that Swenson said in his books. “Now, I wasn’t beginning to end, but you know, managing the board, selecting the board, managing that political nonsense away and, and hiving off the investment process and hiding it from those political machinations.”

So I do think that’s a reference book, but you’re right. It doesn’t, beginning to end. It’s sort of scattered through the two books. Sorry, Rodrigo, to jump in. You had something to say, so I’ll…

[00:17:32]Rodrigo Gordillo: No, I was just going to say that, you know, we’re quants and we sadly, you know, we’ve written a book and we’ve written a bunch of white papers and we get calls from very important organizations that were like, Oh my God, we’re going to call from that. I mean, that’s fantastic. Then we meet that technical person that read the paper, have an amazing conversation and we get down to brass tacks. I’m like, okay, so what, how should we kind of think about getting an allocation in your firms? Like, Oh no, you’re going to have to talk to John over there. He’s the guy, he’s the decision maker. Okay. What’s his background? Oh, he’s just been in the business forever. Does he know anything about quant? No, no, but you know, I’ll put in a good word for you.

And then you’re like, you just have a, have to have a completely different conversation with the two people in that same organization. One that gets it, loves it, gave great feedback. And the other one that you really have to start talking about meat and potatoes. Right? And so the more technically oriented people that we can get in positions of power and decision making positions, the better. And this is what I think your book is good at doing. And hopefully more of those quants end up reading it and asking for promotions from here. It would be fantastic for every one of us.

[00:18:49]Michael Robbins: And your example is not even close to the worst case. If you’re talking to, say, a pension board with bus drivers and firemen on it, or a client who has maybe extensive experience in his field, say, chemistry, and no experience in investing, which is even worse because then they have the position of authority and a feeling that they should understand when they don’t really, and then they may be very defensive.

They might actually be adversarial and you have to win them over psychologically and personality wise before you even get to why this thing should work. And they honestly don’t even know whether you’re telling the truth or not, because it’s just all nonsense to them.

I’ve been in adversarial meetings where either I was defending my strategy, or I was trying to take over the business from another competitor. But the competitor was actually in the room and we had to debate in front of the board about who had the more viable strategy. And the board didn’t know whether, you know, our arguments made sense. My best guess was they were trying to read our faces and see if we could convince each other.

[00:20:07]Adam Butler: This is such a common problem in all technical fields, right, where inevitably at some point there’s a layperson that needs to make a decision. And, you know, it’s actually, I’m on a couple of boards that are related to swimming. And, one of the things we had to do is hire a new swim coach. Well, I’m not a technical person in swimming. I don’t have the skills to hire a new, a new swim coach. So, you know, how should I go about making this decision? Right. But boards. all the time are charged with making decisions in areas that most of the people on boards don’t have any direct technical expertise in.

And I also agree, you know, actually, I want to pull a little bit on this thread. You mentioned this sort of, the chemist and other technical people, obviously that they’ve gone through a lot of school. They know their own domain extremely well. They perceive themselves and in their own field are highly technically astute. What is it about finance that makes it so hard for most other technical domains to connect with intuitively?

[00:21:22]Michael Philbrick: Before we do that, before you answer that, I just want to remind everybody on here, Got to Like, and Share, brought to you by ReSolve Asset Management, And you helping us get a bigger audience helps us get great guests like Michael Robbins here on the show. So make sure you Like and Share.

And by the way, participate in the conversation. I see we’re getting some comments coming in. If you have questions for Michael, hit us up with them. We’ll try and work them in. And again, we’re up to a lot of neat stuff at ReSolve Asset Management. So Like, and Share, check out the content on And, back to you guys.

[00:22:01]Michael Robbins: And while we’re doing shameless plugs,

[00:22:03]Michael Philbrick: Yeah, there it is.

Yeah,,. Michael Robbins.

[00:22:09]Michael Robbins: I’ve got others in the pipe. And if this one doesn’t do well, I won’t be able to get a publishing contract to put the other ones out.

[00:22:17]Michael Philbrick: All right. Now, Adam, back to your question. I, I’m sorry, I forgot what it was.

[00:22:21]Adam Butler: What is it? Yeah. I’m just to remind you, what is different about finance that makes it so hard for other very well educated technical people to connect with intuitively?

[00:22:31]Michael Robbins: I don’t think it’s so much different, actually. I think it’s a similar problem in most technical fields, that it’s not interesting unless you see the benefit from it, right? So, you know, medicine is not that interesting to a lot of people until they’re sick. And investing is not that interesting until you see how much money you can put in your bank account or take out, right?

Alot of people struggle with financial concepts until they start trading and then they get it right away, because once it benefits you, you sharpen up. Who was it? I forget. Someone really famous said that, like he was supposed to be executed in the morning and he mentioned how quickly it sharpens your mind, knowing that you only have 12 hours to live, right? Something similar to that, like floor traders, even options traders, where options is, is incredibly technical, they might not have a lot of education, but they can understand this stuff inside and out, at least intuitively.

And brilliant people, brilliant students at top universities struggle with it. I bet if you stuck them in an options pit for a couple of days, they’d figure it out really quick. And the same is true with finance. It bores people to tears. But if they see how much money they can make, they’ll be really happy to learn it.

And most of it is not that complicated, although some of it is. And a lot of it is a lot more complicated than it looks. And that’s a big point I try to make in the book. Some things are just not that simple, as you probably know, like,  commodities, ETFs. They’re horribly complicated inside and they don’t behave at all like you might think they would if you thought they’d mimic the spot price of the commodity.

Like, VXX is something I mentioned in the book, how it doesn’t track volatility over multiple day periods. In fact, it’s almost guaranteed to lose money over time, over pretty much any time frame greater than a day, just because of the way it’s constructed. Right. The way the vehicle is made, the contract and the way they hedge it at night.

[00:24:44]Rodrigo Gordillo: Well, even simple contracts with Carry like Bitcoin, the Bitcoin tracking futures, right? Oh, we’re going to get exposure to Bitcoin. Maybe, kinda, not necessarily. Depends on timeframe and what’s happening. Right. So yeah…

[00:24:59]Michael Philbrick: And then the construction nuances where you do it all between 4:00 and 4:15 on the on the day, which did cause some consternation for a couple of those of ETFs a few years ago.

[00:25:11]Rodrigo Gordillo: Yeah. Well,

[00:25:12]Adam Butler: Yeah, that’s …

[00:25:13]Rodrigo Gordillo: I mean,

[00:25:14]Michael Robbins: Yeah.


[00:25:14]Adam Butler: I think, I guess where I was going with that was that, um, you know, if you’re an engineer and you’re used to working on civil engineering or mechanical engineering, you’re used to, if you’re going to, if you’re going to turn a dial. You have a very good, you have high confidence in what effect that’s going to have immediately on the process that you’re mediating, right? Whereas in finance, because the signal to noise ratio is so low and because there’s so many dimensions of randomness, that you know, it’s, I find it takes people a very long time. I think many people just never really fully internalize the impact of randomness and luck in financial markets because it really doesn’t impact them in the same way, or at least they don’t pay attention to it in the same way in other dimensions of their lives.

[00:26:09]Michael Robbins: yeah, I think, well, there’s a couple of points there. One is, it’s a big problem in a lot of things, right? Say weight loss, right? You might enjoy cake at, you know at dinner. If you diet over the course of several weeks, you’ll lose weight. People don’t have that patience to wait it through. A lot of people, I think a lot of amateurs who invest think that luck plays too big a role and they don’t put in the work they need to do because like you said, the noise overshadows the signal and they, they just don’t see the true benefit of putting in the effort and playing the odds. It reminds me of an anecdote that Cliff Asness wrote, Cliff from AQR. And, I’m going to butcher it, but he wrote something like he was walking down a hallway in his offices in Greenwich, and he passed by one of his portfolio managers and the portfolio manager was really excited about a stock and said he wanted to overweight it and give it a really excess percentage in the portfolio. And he was really excited. And he wanted to know what Dr. Asness thought about it. And he said, look, I’m a quant, I invest in hundreds, thousands of positions. I don’t even know if I’m long or short it. Right? Because he’s trying to minimize the effect of randomness by creating a lot of bets so that his skill overwhelms it.

And that’s really a big part of the process. I think another thing that relates to what you mentioned is the difference between the natural sciences and investing, and that’s that you can create experiments in many natural sciences, or you can retry the same thing over and over again. You can’t do that in finance.

Every time it’s just a little different in one way or another. You can’t just keep trying it and average it out, a possible exception being, you know, the science of causal inference, machine learning, where they’re trying to create the ability to do that. But they’re relatively limited. right now.

But I must put in the caveat because a friend of mine will get upset if I don’t, medicine isn’t like that either because you can’t just try things on people and, right, there’s the FDA, they, you know, there’s limits to your experiments, at least in this…

[00:28:33]Adam Butler: Well, there’s randomized, double blind, placebo controlled trials, right? Those are the gold standard for medicine. They do that, you know, anytime there’s a new …

[00:28:42]Rodrigo Gordillo: Yeah. But you …

[00:28:43]Michael Philbrick: Contaminated by funding, what gets funded, what …

[00:28:45]Rodrigo Gordillo: But also …

[00:28:47]Michael Philbrick: … what, what, what new…

[00:28:48]Rodrigo Gordillo: …especially in nutritional medicine, you just, you, you can’t force people to like starve to death for you know, a full year. There’s, there’s a bunch of limits that you can’t pull people through.

[00:29:01]Michael Robbins: They …

[00:29:01]Rodrigo Gordillo: … did do that and they like, and they’re not doing it again.

[00:29:06]Michael Robbins: They’re not doing it again.

[00:29:06]Rodrigo Gordillo: They did starve people for 40 days under the guise of, they were very religious, back in, I think it was in the early 1900s. They were very religious. And they said, you know, Jesus fasted for 40 days. So you’re going to fast for 40 days and we’re going to do these experiments on you. Right? So I don’t think that would fly today, but there are limitations in medicine.

You’re right. That experiments, you can’t just run any experiment in a way that you could run experiments on some sort of geological fact. Also, that doesn’t change. There’s no reflexivity in a geological experiment in a way that markets have, right? So that’s one of the things that I’d love to, for you to tell us a little bit more about, and when you’re dealing with trading large amounts of money, right? As a prop trader, when you’re trading a little, a little, you can do, you know, tick data trading and have a lot of numbers that you can lean on in order to get a signal.

But when you’re dealing with large amounts of money and possibly end of day trading, how do you create experiments, experimental designs that actually give you a signal that isn’t hoodwinking you.

[00:30:13]Michael Robbins: Yeah, I think actually the opposite is harder, but to answer your question directly, I don’t want …

[00:30:18]Rodrigo Gordillo: No, no, no. Let’s, Let’s, let’s, touch, let’s touch upon that one later.

[00:30:21]Michael Robbins: Um, yeah, um, we do a lot of experiments with market impact and we, it’s hard to get the data right. And you have to rely on things that have happened already. You can’t just say, well, let’s push buy a billion dollars worth of Apple and see what happens to the market and measure it, because it’s too expensive to make a mistake.

But you can try to infer it. You could try to create microstructure models to try to figure out what the difference is between the impact of large trades and small trades and how that may be affected better, spread it out over time where you, you have the risk of maybe the market drifting away from you rather than being impacted by your large trade.

But I actually think, and also with large trades, concentration risk is a big problem, right? You might actually have to buy a significant percentage of the available supply of whatever it is you’re trying to buy, which is another problem with large trades. But I had a good friend who I traded with for a long time and he traded fine with large trades. He traded hundreds of millions of dollars at a clip. And then when he retired and tried to trade his own money, that’s when he had problems. Yeah, the whole psychology of losing his own money really affected him, and he couldn’t trade small lots.

[00:31:43]Rodrigo Gordillo: We’re robots. All right. Yeah, they’re human too, I guess. They need psychologists just like Axe Capital. Every quant shop needs an Axe Capital type of psychologist on the docket. I get it.

[00:31:59]Michael Philbrick: Yeah, I wanted, I do want to get to like the relationship between quant investing and data science, because I think the data science side is very, very different and a little bit more new and upcoming, but I don’t want to, I don’t want to jump on, I think you have a train of thought Adam, but I don’t know if that’s, if that’s dovetailing for you or not.

[00:32:17]Adam Butler: Well, no, I wanted to sort of stay on because I really want to, I want to pull on this thread of, because we don’t get quants who work at or manage, you know, large institutions on very often. Right. And our experience with dealing with many large institutions is that, you know, as we sort of discussed, the management team is, I’ll never forget, I remember sort of going to many, many conferences, sort of five, six, seven, eight years ago. And there’d be these junior analysts in the crowd. And then there’d be the senior, the C-suite decision makers of the crowd. And you’d have an expert up discussing some, you know, relatively sophisticated or technical portfolio oriented subject matter or, you know, about risk premium and getting into the weeds and the senior decision makers, you know, sitting back, clearly not watching, not caring. They don’t have a clue what the guy is talking about, and the junior analysts all engrossed, asking questions, challenging the speaker, whatever, and thinking, um, this seems so backward, you know. And I’m just wondering whether or not your observation, the decision makers at these large institutions are getting more sophisticated. Are they embracing more quantitative or systematic decision making in their workflow and in their process? And, you know, how can institutions think more effectively about using quantitative methods? What are the low hanging fruit that are often missed?

[00:33:59]Michael Robbins: Yeah, I think decision science is definitely having a big effect on sophisticated institutions like hedge funds and banks. They’re all over it. Like pension funds, depending on, you know, which one you pick may or may not be. I imagine some of the state pension funds really don’t know the first thing about it. Some do, some of the more sophisticated funds like Texas Teachers, or Canadian funds are probably just all over the stuff. But like, when I was a young trader, I was really distraught because I’d spend all night or all week of sleepless nights, coming up with an answer for my boss. And in 20 seconds, he just say, no, that’s wrong.

Fix it. Right? And because he just knew what the answer should look like in his head, he didn’t need to calculate it. And I think we’re still at that stage, mostly for investing, that intuition and experience are still far more valuable than what we can calculate. I mean, there are certainly opportunities on both sides, but machines aren’t ready to overtake the world just yet.

They can take advantage in scale and complexity and speed. But for the big things, like when you read the news about, like this … takeover and, you know, the stuff about Twitter and like all the, these big deals, you don’t see a lot of data science in those stories. So we’re still waiting for the singularity where computers can outthink people. Up until then, I think the senior people will have quite a bit of input.

[00:35:43]Michael Philbrick: But then computers will be out-thinking computers. I mean, someone has to be at the helm of the computer. And now we got a computer war.

[00:35:49]Michael Robbins: Well, absolutely right. Uh, it’s, you may remember, there was an article maybe a month or two ago about, they were trying to use computers to win that table game Go. You know, the board game and the way they did it was, the guy that won didn’t ask the computer how to win the game. He asked the computer what mistakes the other person was making. So it was more about warfare than winning. Right? And it’s a good point because if you think about machine learning and high frequency trading, high frequency trading, in my understanding, isn’t that sophisticated from a trading standpoint. Yeah, they’re really fast, they have great communication, they’re co-located, they use special hardware and all that.

But they’re not coming up with these really complicated multi-leg trades, with implied values and latency and stuff like that. They’re just doing things really fast. And… I think at least for a while, computers aren’t going to be quite that smart. They’re just going to be able to do things fast and with a lot of data.

And if you’re a thoughtful person or if you have a program that has a different time frame that’s a little more thoughtful than maybe the big computer is trying to think very quickly or handle Exoscale data, then you can find chinks in their armor and scrape out a little money for yourself. And where that really works is if you can have a niche, like maybe Goldman Sachs leaves money on, a lot of money on the table because they’re interested in the really big trades.

It doesn’t pay for them to take all the money off the table. Not yet. Not until they get better at, you know, their data science. And for a while, maybe they’ll leave a few coins there, which might be more than enough for the middle market guys who only need to make a couple of million a year. So I think you’re absolutely right that warfare aspect, that spy versus spy is where it’s going to be at for the next 5-10 years.

[00:37:49]Michael Philbrick: Always invert.

[00:37:51]Rodrigo Gordillo: So, Michael, just back on the topic of what institutions are, you said that there are certain institutions that are really accepting of quantitative methods, but quant can mean so many things. And I think it really, it really does depend on what the quant is pitching, whether it’s largely acceptable in an institution or not. What do you think are the areas of quant that are being used right now in larger organizations? Like what type of styles, what type of strategies are the ones that are most appealing?

[00:38:25]Michael Robbins: Yeah, I’m not a big expert on this. But what I seem to see is that when things become really technical, to the point where people really can’t do them well, then these institutions are forced to rely on technology. So if it’s something like NLP or data scraping or high frequency trading, there’s no way a firm can get away from relying on technocrats, but when it comes to decision making at firms with a smaller headcount, meaning not like big banks, which have, you know, tens of thousands of people. People still trust their intuition and, you know, rely on human beings to validate decisions. And there’s a good reason for that. It’s, the example I like to use is, there are tons of car accidents every year, right? It’s, if not the leading cause of death in most geographies, the second, and it’s obviously a flawed system where people drive cars, but it doesn’t get in the news because somebody driving into a tree as inexplicable as that is, is people understand it, but you have one robot driven car that drives off a cliff and it’s in all the papers.

[00:39:45]Rodrigo Gordillo: And every politician’s on it, right?

[00:39:49]Michael Robbins: Everybody’s on it, right? And so that’s the problem. If you run a company that’s beholden to investors or shareholders, you don’t want your company to drive off a cliff. It’s okay to drive into a thousand trees. You can explain that away, that happens, but you don’t want to drive off a cliff.

So, unless you have a special relationship, if you are, say, a high risk hedge fund who has investors that understand the risk you’re taking, people are trying to avoid that existential problem of making a mistake that may not be so bad, but is irrecoverable. And I think we’re still at that point right now.

[00:40:34]Adam Butler: So, I think you’re saying that quantitative methods have a very high, we have a very high confidence that quantitative methods are going to lead to better answers most of the time, but you need humans at the helm to supervise and make sure that nothing unusual or excessively, you know, risk taking, is going on behind the scenes. Is that where you’re going? I’m just…

[00:41:10]Michael Robbins: Well, that would be nice, and I’m sure it’s possible if the people at the helm are exceptionally self-disciplined. But I’m sure you read about a month ago, there’s a paper where they did a study with pathologists, and they found out that the pathologists and the AI program that was identifying pathologies, were good at different things.

And so they came up with a theory that was very close to what you just said, that what if the people just work with the machines, then they’d take the best of both and be better than each other. And in fact, the opposite happened. People trusted their intuition in exactly the circumstances where they should have trusted the machines. And they ended up with a worse result than if they had either done it themselves or let the machines do it. So these behavioral biases that we’re all subject to, one of the most insidious features of them is that you can know they exist. You could be trying to guard yourself against them and they still get you.

[00:42:17]Rodrigo Gordillo: Okay.

[00:42:17]Michael Robbins: And …

[00:42:19]Michael Philbrick: To know they exist does not immunize you.

[00:42:22]Michael Robbins: Not at all.

[00:42:23]Michael Philbrick: Greenblatt, Joel Greenblatt has features…

[00:42:26]Adam Butler: You should share that story. That’s a great story too.

[00:42:28]Michael Philbrick: Yeah. Um, I, I’m going to bastardize the numbers here, but Greenblatt, you know, the, what’s that little book of investing? So…

[00:42:35]Adam Butler: Little Book That Beats the Market, Herb Greenblatt. Yeah.

[00:42:38]Michael Philbrick: Yep. Here’s the trades you can, we can execute them for you discretionarily or you can choose. Two pools of people executed discretionarily, mindlessly, Groundhog Day, every trade all the time in the right size, outperform the S&P substantially. Person chooses the trades that they want, strategy underperforms the market S&P substantially. So to your point, there’s a few studies out there that…

[00:43:09]Adam Butler: But that invokes the question and you know, you spend time in the book, a lot of it on governance, which I thought was great, and it really sets the stage well for the rest of the chapters. It’s actually really, it’s structured really intuitively for, especially, I think, for non-quants. If we know that systematic thinking and largely leaning on machines for decision making on average, is considerably advantageous from a governance standpoint, why don’t we lean into that when, you know, everybody is standing back thinking about the problem more abstractly, right? You’re not in the heat of battle, which is obviously when it’s very hard to abandon your intuitions. It’s, you know, that’s really when your biases take hold of you and it gets very hard to stay disciplined. But in when everything’s calm and we’re, you know, there’s a process to set your governance standards, why are we leaning more in that direction at that point and setting up guardrails and rules to make it difficult for those behavioral biases to manifest in negative ways?

[00:44:26]Michael Robbins: Yeah. I’m not a psychologist, but the, and before I get into the answer, thank you very much for what you said about the book. I wish I had you to talk to my publisher when I was editing it. Everybody fought me on the structure of the book. Um, but, uh, my, my initial thought about what you said. Um, it’s about how people think, right?

People latch on to the most, the examples that are most relevant in their head and not the underlying, plethora of real. So that one time the ER surgeon identified the rare disease, you know, one out of a million times, that’s what sticks in his head. And he’s always thinking about that when he was a hero.

Every time somebody gets wheeled into the emergency room, even though 99 percent of the time you know, it’s a horse and not a zebra, right? And so your brain just can’t let go of that concept that you can add value with your vast experience in education. And yeah, it’s true in those rare circumstances, but it’s not true most of the time.

Most of the time it is a horse. And a lot of the time you think it’s a zebra and it’s that in between time, the lot, but not the 99th percentile that you’re going to be wrong and you don’t want to give that up. It has to do with ego, availability, bias, emotions, right? When you remember that one time you were a hero, it makes you feel good.

[00:46:07]Michael Philbrick: Smidge of overconfidence.

[00:46:10]Michael Robbins: A lot of overconfidence for sure, but it’s also how people perceive truth. Right? The way we remember whether we’re right or wrong is whether it makes us feel good or not. And so that surgeon remembers the one time he was a hero and it makes him feel great. And so he thinks about that a lot.

Right. And it seems truer and truer every time he thinks about it. It reinforces the pathways in his mind. And that becomes the memory that appears most. And all those times he was wrong, you know, they’re either painful or not memorable. He doesn’t think about them that much. And the times that the rules were right, when you didn’t even have to make a decision, he doesn’t think about it at all.

And, I think that’s a big reason why it’s so difficult to have that middle ground between being purely systematic and being purely intuitive. It’s a really hard tightrope to walk.

[00:47:09]Michael Philbrick: To be fair though, we, we should be sort of thankful for that as quants because that is the very bias that we tend to want to or hope to exploit, right? That the fact that it leaves the door open for more efficient assembly, more efficient extraction of factors on a more consistent basis with a larger sample size that you know, starts to remove luck from the equation. So we, we do, you know, as much as I would love a broader adoption at the same time, I, you know, if I get everybody adopted on this, then maybe my edge goes away because those frailties behaviorally are not left to be harvested by those of us who are using more quantitative methods. So it’s, it’s kind of a…

[00:47:54]Rodrigo Gordillo: So, you know, it’s funny, I was listening to a podcast today and they were talking about Bridgewater and the way they described Bridgewater was hilarious. It was, Bridgewater is a quantitative investment firm that makes money to pay for 1,500 employees to sit down at a table and talk themselves out of intervening with the system that’s worked. That’s basically all they do. I think this, I’m like, nope. No, there’s 10 people on this table that are going to talk you out of fucking it up, right? That’s an interesting thing, right? You have something that’s working and then you want to find the edge case, right Mike? And you have to have a big group of big brained individuals actually saying, Yeah, actually that one time this year, that we’re going to change. But the other 50 ideas, we’re not going to touch. It was good chatting with you though, right?

[00:48:49]Michael Philbrick: Well, in keeping track of all those false positives where you said I want to do something, you didn’t really document it and it didn’t need to be done. You’ve forgotten about that. That one’s not even going on the tick sheet for qualification of, you know, looking at how accurate you might be.

[00:49:06]Michael Robbins: Yeah.

[00:49:06]Michael Philbrick: I think IRImpossible has got a neat comment too.

Often the machine’s reasoning isn’t well represented or articulated to the human or isn’t presented at all. I think this is absolutely pervasive across a good quantitative strategy. Are you kidding me? The multi-dimensionality of making a decision on an individual security or asset class in the context of the diversity of a portfolio, the balance of the portfolio, the managing of risk of the portfolio, all of those are contributing to a multi-layered multi-faceted decision. Of course, there’s not going to be any intuition to that.

This is a, there’s a 20 by 20 matrix of influences and probability estimates that are giving you some sort of edge. Yes, you are not going to have a lot of intuition. I think this is one of the things we do, we try to do well is try to create intuition, right? Stock/bond portfolio. Stocks are up today. Bonds are down. You kind of know where your portfolio lies. A portfolio of 85 futures contracts, long and short, across everything from softs to metals to gold. You, you’re not going to have any intuition of what’s going on. Yeah. So I think you’re right.

Uh, IRImpossible. There has to be this, this increase of a higher level of reporting, a higher level of interaction to develop the intuition that you might have with the system that you’re employing because at the end of the day, if you falter, you don’t falter when you’re up a lot. You falter when you’re down a lot, thereby crystallizing the risk without receiving the return. This is the function of how we, how we quit these types of things. And then, later on that, the political discussions, which are, I’m trying to get that Michael Robbins and I want to get his strategy over to mine. Is he in drawdown yet? Because I’m going to put that pain right on the board. I mean, this is, you know, as Machiavellian as it gets.

So yeah, let’s get Michael’s position out. Asness and his private equity guys. Remember the story of him? The private equity guys are like, look, no vol. And he’s like, you’re not, you don’t have a price. He’s over-loving it, because he’s doing well on the quant side where they’re striking a NAV, but they’re after his money. They’re like, give me the allocation. What are you doing over here? I mean, what a …

[00:51:32]Michael Robbins: Don’t forget the Hunt Brothers.

[00:51:33]Michael Philbrick: … spaghetti.

[00:51:34]Rodrigo Gordillo: Preach, Mike. I mean, creating a sense of intuition, quant, is super important. But I do wonder sometimes if you’re just creating a good story for us to feel good about, or for the investor to feel good about.

[00:51:48]Michael Robbins: Well,

[00:51:49]Michael Philbrick: Of course we are.

[00:51:50]Rodrigo Gordillo: And that’s all it is.

[00:51:53]Adam Butler: Okay.

[00:51:54]Michael Robbins: But uh, you brought up some points I’d like to comment on if I can, quickly, because they’re really good.

[00:51:58]Michael Philbrick: Please.

[00:51:59]Michael Robbins: The story you were referencing, Matt Levine from Bloomberg Opinion talks about something like that, very much like that a lot. He’s one of my favorite authors, and he has a slightly different interpretation of it. He says that the people at Bridgewater have this great machine making all this money and their culture is to argue with each other so much that they don’t interfere with the machine. Right. That the, their biggest benefit of their culture is it keeps them away from, you know, the dials.

[00:52:30]Michael Philbrick: Hey, put, put one mark for resolving that.

[00:52:33]Michael Robbins: Yeah.

[00:52:34]Rodrigo Gordillo: Amen.

[00:52:36]Adam Butler:

[00:52:37]Michael Robbins: Another story that I thought of while you were talking. It was a great paper I wrote, which is a play on words from Kahneman’s book. Trading Fast and Slow is the name of the paper. And they did this study of, I think, money, professional money managers, a lot of them, and they determined pretty conclusively that they think a lot about what trades they want to get into when they create their sales pitch.

But when they have a new thing that they want to buy, they don’t think really much at all about what they sell to raise the money to buy the new thing. And they effectively leave about 2 percent on the table by doing that. And I’ve, I’ve worked at a lot of big money management firms and I’ve asked a lot of people and everyone that I asked agreed that, yeah, that’s absolutely true.

You know, when we’re trying to sell something to a client, we just want to raise some money to finance the deal. And, you know, maybe you buy something that’s up a lot to take profits or buy something that’s down to, you know, stop losses, but we didn’t, we don’t do a lot of analysis to try to decide, is this the opportune, perfect time to sell, and maybe we should wait on buying that new thing because it’s not the best time to sell the old thing.

[00:53:50]Rodrigo Gordillo: Yeah. I believe that’s the academic term for that is a shiny new object.

[00:53:56]Michael Robbins: Shiny new thing. Yeah,

[00:53:57]Rodrigo Gordillo: Well documented, but it’s true. I mean, you want it so badly that you forget that it’s just as valuable to be thoughtful about the exit of old strategies that you think don’t work as it is to invest or start investing in the new strategies, for sure. Now it’s, it’s, it’s an end to end problem that needs to be solved.

[00:54:18]Michael Robbins: Yeah, but serious people do performance attribution on themselves and they have alpha capture programs and things, but most investors don’t. And I’ve never seen an RIA that…

[00:54:31]Adam Butler: Silence.

[00:54:41]Michael Robbins: …benefit to be right to figure out what they’re doing right, and what they’re doing wrong, and improve the things they’re doing wrong, or stop doing them, right? You just don’t…

[00:54:50]Michael Philbrick: …think there’s an objective function there too. That’s a little different. Right? And it’s sort of related to career risk, right? It’s sort of similar to the asset manager. You say, well, here’s a better mousetrap. It’s quantitative. Well, is anybody else doing it? How far away from the benchmark is it going to get me?

So the asset manager has the problem. The RIA has the problem of being fired, you know, suggesting to diversify and to manage futures 10 years ago, as an example. And, and sitting with that drag on the portfolio for a decade while preaching diversification gets you less clients, keeps you less clients.

[00:55:26]Michael Robbins: Although a lot of advisors do that. But, what I was talking about was actually studying and attributing the different parts of their performance. Right? So a lot of them keep track of their performance. They have to, they have to tell their clients how much money they’re making. They may play games with benchmarks and things like that, depending on how, you know, how much they want to play games with their clients, but to do like an in depth attribution to really be serious about, hey, am I good?

Being long or short, am I good at picking energy stocks or utility stocks, right? Am I good at, you know, when momentum regimes are in play or mean reverting regimes, right? Just really picking it apart. That’s a really technical thing. That’s something that, you know, most investors don’t get into.

[00:56:18]Rodrigo Gordillo: Right. And this is…

[00:56:20]Adam Butler: When we, when we got going, oh, sorry, close the loop on that if you want, Rodrigo.

[00:56:23]Rodrigo Gordillo: Done.

[00:56:25]Adam Butler: Okay. I was going to say Mike was, he sort of led with a question about communicating quant concepts and the merits of quantitative approaches and quantitative strategies to lay people or to investment professionals who are not, don’t have a background in quantitative methods. Any insights on best practices there?

[00:56:53]Michael Robbins: I’m, you know, I’m not terrible at it, but I’m not that good at teaching it because I don’t really understand it that well. I will say that, my faith in people to understand complex problems really took a hit during COVID and people seemed a lot, well, they really didn’t seem to be able to grasp some concepts and it seems really hard to communicate something to somebody that they don’t want to hear, thank you. And if they’re not receptive, trying to explain something as complicated as like an LLM is probably impossible. It might be nearly impossible under the best of circumstances, but oftentimes when you’re trying to communicate to someone, they don’t want to hear what you’re saying anyway. And it really comes down to trust, which is why I talk a lot about in the book about building business plans and experiments so you can show people results and the best they can say is, I think you’re lying to me.

All right. Well, if you think you’re lying, I’m lying to you. There’s nothing I can say to change that, but assuming I’m not lying to you, I have all this proof, right? And I can show you all this proof. And if I think you’re lying, you don’t even have that, right? So, we’re really at a standstill, at least…

[00:58:15]Adam Butler: So what is, what does proof look like to quantitatively oriented though?

[00:58:20]Michael Robbins: So that’s a big problem, right? Cause that’s the whole thing behind back-tests and the, and the criticism of back-tests, lots of people abuse them. And so they get a really bad name, but they’re actually a great tool if you use them right and if you’re honest. And it’s like anything else. If you’re going to abuse it, nobody’s going to believe you, no matter how good the tool is.

But if you use it correctly, you can provide a lot of good results. And I’ve been in heated discussions about back-tests, and you can defend them to someone with an open mind. They say, you know, I don’t believe your back-test. You say, oh, why? Well, you know, a lot of back-tests, you know, overfit. Yeah. But look at these ways that I tried to prevent overfitting.

I did this cross validation. I tested out a sample. I, you know, I did all these things. I tested these hypothetical situations using synthetic data, and then you get into how you made the data. And I said, Oh yeah, yeah. But, but you, you know, a lot of times the transaction costs are not really realistic.

Look at how I did those, right? And you can get into the weeds and you can refute their concerns. You may not change hearts and minds, but you can certainly make a good argument where they don’t really have a lot to push back on. It really comes down to emotions. And that’s kind of the best I think you can do if somebody doesn’t want to hear what you have to say.

[00:59:48]Michael Philbrick: What question to ask. Is there anything I could say that would change your mind? And if someone says, nope, there’s nothing you could say, you’re like, okay, well that’s, you’re closed for business. That’s fine.

[00:59:57]Rodrigo Gordillo: So, so …

[00:59:58]Michael Robbins: …you think? And I’ve …

[00:59:58]Michael Philbrick: …nothing I …

[00:59:59]Rodrigo Gordillo: Let me just take this, this time to, kind of give a plug to our Chief Investment Officer and head quant, Andrew Butler. We did a podcast with him a year ago that actually delved deep into all these things, process, how to think about back-testing, how to make sure you’re not fooling yourself and you’re the easiest one to fool.

And you know, there’s a whole, it’s super important as we head into this world of machine learning and LLM, it’s becoming really, really popular that. Allocators understand the difference between a full in-sample back-test and an out-of-sample back-test. Like, simple things like that. Just ask that. And then learn, so that you know at the very least how to weed out the quants that don’t even know they’re fooling themselves. And there are a lot, right? There are a lot of them, especially putting out ETFs and mutual funds, that don’t understand a proper experimental design, using the proper scientific method to make sure that you’re doing it right. And if you want to listen to that podcast, you can go to www. and just type in Andrew Butler. And I think the podcast is called ReSolve’s Own Dr. Andrew Butler, Integrating Prediction with Optimization. So do check that out. Also, if you guys are still sticking around, please do Like, and Subscribe.

Mike’s a great guest, you know. I think we’re, we’ve learned a lot from reading his book, Quantitative Asset Management. I think you guys, uh, will too. So please take a, go to his website and download it, read it, I’m sure you’ll do one. Do you have an audio book? Cause I’m sure that will.

[01:01:35]Michael Robbins: I’m making videos and I’m posting them on…

[01:01:37]Rodrigo Gordillo: Video. Video on the website, so go check that out. But let’s talk a little bit about…

[01:01:42]Adam Butler: Well, what’s that tool that reads books for you? Like  reads PDFs? I forget now that we were talking about this week. Speech…

[01:01:48]Rodrigo Gordillo: … only that, Speechify now… I forgot to give it a plug. Not getting paid for this one. Speechify now summarizes… Like you can actually, you send it in, it reads it to you. And then it also summarizes and reads you a summary in case you want to, you know, not waste an hour listening to a long form white paper. So they’re, they’re getting, they’re upping their game. It’s pretty cool.

[01:02:11]Michael Philbrick: Strong, strong word.

[01:02:13]Rodrigo Gordillo: But yeah, can we dig a little bit on proper back-testing? I mean, how do you think about that problem?

[01:02:20]Michael Robbins: Yeah. There’s a lot of nuances to it. And unless you have a really fancy computer, which a lot of people do, but most amateurs and even professional traders don’t, you got to pick and choose what’s important to your back-test, which makes it. A lot more of an art than a science. Uh, you could try to include realistic fees and costs, taxes if they apply, market impact, you know, all sorts of things like that, but then it, the back-tester gets so bogged down that it becomes ineffective. And then there are, you know, lots of other interesting things that you can add in. Marco Lopez de Prada talks about penalizing your back-test for the number of back-tests you do, right? Because the more you do, the more likely you are to stumble on something that it works by chance and not in reality, right? And there are just so many interesting ways to enhance your back-tests. And, you mentioned it earlier in this podcast, there’s so many decisions you have to make when you make a quantitative model. People think of it as a science, but really every little thing you do involves a decision that’s scientific. You can’t hyper parameterize.

Everything, you have to just try a few things. The space is too large and multidimensional and being able to do that is really tricky as well. So, yeah, it’s really kind of an art, but what I try to get into in the book and what you describe as well in your podcast is all the little things that you can do to improve your back-test to make it better, right?

And importantly, your back-test can overcome some of the oversimplifications you make in the other parts of your design process, in your factor research, in your optimization, right? All these other methods are much more computationally intense. They take a lot more time and have to be simplified a lot more than the back-test does.

So you can use the back test kind of as a sanity check and say, did my other model? And that’s a useful thing too, because you might, for instance, use the triple barrier method as an objective, right? Where you have a time terminal where you say, if I’m in a trade for a certain amount of time, I’ll exit it, but you might also have an upper barrier and a lower barrier for a “take profit” and a “take loss”.

And that might be a very oversimplified example of what your investment process is like. So you may have that in part of your research, but then in your back-test, you may put in something much more complex to better test your theory. And that all depends on how you work, right? Maybe you don’t do trades unless you have a board meeting.

Maybe you don’t have a board meeting on holidays, or you can call an emergency board meeting if volatility spikes, or what have you. The more that you can build into your back-tests, the more ammunition you have to keep money in your strategy when things start going south and people want to pull the plug.

So there’s a lot of nuance there. And there’s a lot of really fun things you can do. And what’s great about it, which goes to some things that we were talking about earlier today, is it’s different for every institution and for every strategy, right? Every institution has a different fee structure. They may have different tax structure., They may have dark pools where they can cross trades and pay almost no fees, or maybe they have to pay full commissions, right? All sorts of different things that affect your back-test. So that means they have to hire a quant to do a new back-test. They can’t just buy one off the shelf, right?

And that’s great for us. It just builds a lot of opportunity, a lot of friction to take advantage of, and a lot of nuance and dials to turn to make money for that particular institution. Because you don’t have to necessarily beat the market, if you could just beat the existing strategy and improve from what they have then.

[01:06:38]Michael Philbrick: …that. And, I think just to clarify.

[01:06:40]Adam Butler: Good, sorry, yeah, I’m just going to make a statement and then you could, but I think back-testing is a really good example of where you were saying earlier that experience has a really large role, right? Like, head explodes with a number of dimensions of the kinds of decisions that you can make.

And you get into these discussions with people who just don’t have that level of experience with the data or the methods and it’s really hard to connect the dots. There’s not very good language for the intuitions that you gain from that kind of experience about how you use these two assets and you, with this process, and it looks like it was really good, but there are special properties about these assets that actually made it look really good, but if you, and it was, it was unique to this period and it’s not going to repeat, but you’re sort of deceiving yourself

you know, so, but…

[01:07:43]Rodrigo Gordillo: …give a concrete…

[01:07:44]Adam Butler: …impossible to have like,…

[01:07:46]Rodrigo Gordillo: …what you’re talking about…

[01:07:48]Adam Butler: Well, yeah, so,…

[01:07:50]Michael Robbins: Sounds like you know.

[01:07:50]Adam Butler: Example, as an example, there was an article about replicating Trend, the CTA Trend Index with four assets, right? So, ES, S&P, Treasury bonds, CL, so, oil, and I think the euro. And the author demonstrated that this, that, you know, beat the CTA index over the Middle of 2008 through the middle of 2023. Now the CTA index goes all the way back to 2000, right? So immediately you’re sort of scratching your head. Like, why, why’d you only go back to 2008? And it turns…

[01:08:32]Michael Robbins: That’s always a flag.

[01:08:34]Adam Butler: Sure. But of course, also we know what was the best performing asset, not by a little, but by like a massive amount over the mid to late 2008 period through mid 2023. Like the S&P outperformed everything by like 3, 4, 5X, right? And so having an extremely high, an asset with an extremely high mean as one of your assets to select is going to vastly skew your performance

[01:09:02]Rodrigo Gordillo: And fools you.

[01:09:03]Michael Philbrick: Well, as your only, as your only equity…

[01:09:07]Adam Butler: Equity in this? Exactly.

[01:09:09]Michael Philbrick: One, you just pick the best one. You didn’t pick the Italian Borsa.

[01:09:14]Adam Butler: Right.

[01:09:14]Michael Philbrick: Didn’t, you didn’t pick a number of them where there’s the back and forth and the switching out because there’s a number of them that may be coming into the lead. No, you just pick…

[01:09:24]Rodrigo Gordillo: So what is the one that’s…

[01:09:26]Michael Philbrick: And you call…

[01:09:27]Adam Butler: But it seems super intuitive. Like ES and TY or, you know, CL, these are like the biggest markets. So, yeah, I mean, of course it makes sense. But you see, you don’t really, it’s not, it’s not easy to see the bias that you’ve embedded. You think you’re actually being …

[01:09:40]Rodrigo Gordillo: Well, my favorite one…

[01:09:42]Adam Butler: And so it, this is, it’s…

[01:09:43]Rodrigo Gordillo: My favorite one was when we wrote the Adaptive Asset Allocation paper back in 2012. And of course, you know, this was a long, flat kind of momentum driven, you had, you know, global equities, domestic equities, bonds, commodities, gold, something like that, just a random set of diversified assets, right? And it kind of went viral and a lot of people tried to improve on it. And I remember one group came and said, well, look how well I’ve done. Look how much I’ve improved on your models. And they had, it was, it was fantastic. They had done an amazing job at picking the NASDAQ as the equity position. They took out all the non-performing equity mandates.

And then, what’s more annoying is that five years later, they claim that out of sample, their approach to adaptive asset allocation continued to outperform, proving at a sample walk forward that it is still a better system without seeing the bias right then in front of them, right? And this is how, this is just a super obvious example to a lot of us. But it’s not that, and there’s so much that’s not even super obvious to us that we are still learning as we go through it. So it is, it is a tough and nuanced process.

[01:11:08]Michael Robbins: It’s…

[01:11:09]Michael Philbrick: I just wanted to come back to Michael, what you said earlier about a better back-test. A better back-test does not mean a higher Sharpe ratio. It does not mean a higher return. It means a back-test that is more reflective of what happens out of sample. The character is more in alignment with what happens out of sample, not better in a Sharpe ratio or better in a mean return. It better, most bestly reflects, I don’t know,

[01:11:36]Rodrigo Gordillo: That’s a good one. That’s a good mic. I’m writing

[01:11:38]Michael Philbrick: Woo. I got to go back to English class. Okay. Better, better reflects the out of character, out of sample character. Anyway, over to you.

[01:11:47]Michael Robbins: An example that I was thinking about when you were talking is a little more subtle, but really in the front of my mind, and has been for a very long time. A lot of people write about it, but Aaron Brown wrote a whole book about wrong way risk, right? And right way risk. And I remember when the Russians defaulted on their bonds, it was like 1998, I think, and I was in a basis trade.

And the trade was amazing, right? It was a layup. It was, I mean, it wasn’t a layup until you put it on, but it was an arbitrage and we were guaranteed to make money, but the company was bleeding from their equity department and they needed cash for financing. And so we had to get out of our trade to generate cash for them to finance their overnight positions, right?

Because Treasuries and futures, you know, it’s settled on different days. And so you need to finance overnight and they couldn’t give us the cash overnight. It was just a gap risk. And the same thing’s true. Say, if you have shorts, you could back-test a position with shorts in it. But then when you actually run the position, the bank pulls them from you, right? And so your short position…

[01:13:01]Adam Butler: Short equities. Yes.

[01:13:02]Michael Robbins: Yeah, short equities, right? And so exactly when your strategy is supposed to make the most money is when you can’t execute it, right? And that’s something. If you never experienced it before, you wouldn’t know to put in the back-test. It’s a great, this great short book. It’s amazing.

Or, you know, I make money when everybody else is losing money. That’s a great thing. It’s not a great thing if you need those other people to finance your trade. So there are a lot of subtleties that an experienced person will build into a back test

[01:13:35]Rodrigo Gordillo: The known unknowns, right? These are like the known

[01:13:39]Michael Robbins: Known, well, they’re known if you’ve gone through them.

[01:13:43]Michael Philbrick: Well, remember the guys in Waterloo where they had solved the market and we’re like, well, how did you do position sizing? And how did you calculate the margin for all the…

[01:13:52]Adam Butler: These are PhDs in bioinformatics, right? Extreme, sophisticated, technical people who had never operated in markets before, but claimed to have solved the market. Sorry, Mike. Anyway, just…

[01:14:05]Michael Philbrick: Yeah. No. And really we’re like, well, how are you calculating positions? Oh, we just, the inverse of the current, you know, margin rates or something like that. And I’m like, just the current ones, not the ones from 20 years ago?

[01:14:16]Adam Butler: …know, they change, right?

[01:14:18]Michael Philbrick: They tripled it overnight, not the one where they were the gold and silver. They, you know, it was going through a parabolic spike, VIX, you had to pay one, one. Like, you didn’t have access. You could not put those trades on in the magnitude that you’re suggesting. You didn’t have enough capital. Did you account for any of that?

[01:14:37]Rodrigo Gordillo: Yeah, they solved their portfolio into an 80 percent drawdown in two months. So…

[01:14:42]Michael Philbrick: Typically what happened?

[01:14:43]Rodrigo Gordillo: Yeah, I mean, these are the things, the smartest people, the most technical people, people straight out of Google and machine learning, never touched the markets, not understanding the subtleties of a proper experimental design that takes into account all the all the experience that one has in trading markets themselves, right?

[01:15:04]Michael Robbins: An example I bring up in class all the time is, even from the very beginning, you say, well, what do you want to do? I want to make a strategy that beats the market. Well, how do you define the 30 day rolling returns. Uh, do you have a triple barrier, right? Like that’s a subtle question, even without getting into what you invest in or how, or what the timing is or the rebalancing or what your finances. Yeah. I mean, there’s a million things.

[01:15:30]Rodrigo Gordillo: I didn’t even think of …

[01:15:31]Michael Robbins: …know…

[01:15:31]Rodrigo Gordillo: You have students who think they’re going to be so filthy rich and solve their market with Sharpe ratios of 4. I can’t even imagine how many back-tests you get that are just, like, that get them so excited, they probably feel like they’ve done something nobody else has done, only to be faced with reality, right? Or maybe, maybe the worst thing is that they turn it on and then you’re…

[01:15:55]Adam Butler: We’ve all …

[01:15:55]Michael Philbrick: We’ve all fallen for that on the journey. You get that beaten out of you. The other thing…

[01:16:01]Rodrigo Gordillo: The one that…

[01:16:02]Michael Philbrick: Which market?

[01:16:02]Rodrigo Gordillo: …months, then it’s like then you’re…

[01:16:05]Michael Philbrick: …market do you want to…

[01:16:06]Adam Butler: Now you’re really dangerous.

[01:16:08]Michael Philbrick: Yeah. Right. So, the market that you select to be the one that you beat, well, how did you make that decision? How did you make the decision that it was this market or that market? Uh, I know we penned a piece a while back on, you know, emerging market managers and US equity managers, and just looking at the dispersion and you could have been top percentile emerging market manager, you couldn’t even scratch the balls of the worst US equity manager. Like, and who made that selection? Like you couldn’t even reach a scrotum. So I mean, just by, you know, oh, I picked this market bias, bias, bias, bias. Anyway,

[01:16:47]Michael Robbins: And that’s a lot of what I try to get across in the book. And it’s a positive and a negative. I try to open people’s eyes to all these things that they might not have thought about. And even if you’re a professional who is very good at a specific market, maybe you haven’t thought about AT1’s or, you know, some other esoteric investment or whatever.

I just try to bring up as many things as possible to give people food for thought to kind of burst that ego a little bit. So they can actually make money and not lose it as soon as they go live. Because you’ve all seen those charts of in- and out-of-sample, you know, live trading where they do great and then they start going live and it completely goes…

[01:17:29]Rodrigo Gordillo: Or they flatline, right? They just end up …

[01:17:31]Adam Butler: The key is to do the in- and out-of-sample as your back test.

[01:17:34]Rodrigo Gordillo: Yeah, they did do a new back-test, yeah. It’s like, I fixed the problem, let’s try it again. Nope, same thing. Now I’ve got it. This is, the third time’s the charm. So, I mean, the average edge of any system that actually has an edge is very small, right?

It’s almost a coin toss. So, I mean, how do you educate your students into actually creating robust quantitative models, understanding that limitation?

[01:18:07]Michael Robbins: Well, there are different kinds and, counterfactual to what you just mentioned is for instance, arbitrage, so there could be arbitrages that the big banks, at least for now, it’s not worth their effort to get into. It’s the, the value versus the time and talent it takes to extract the money is low, but for somebody, maybe a midsize bank, who’s pretty smart and knows how to do it, can extract money with a near perfect record.

It just takes a lot of energy and isn’t worth it to most of his competition. That happened to me when I was trading the Treasury bond basis in the nineties. The big guys were going after the big trades. They weren’t looking to make a million dollars. They wouldn’t get out of bed for a million dollars, but that was fine for me. I’ll take a million dollars profit, you know, and I’ll work all day for that. I don’t care, right? And then the smaller guys, the amateurs, they just didn’t know how to calculate it, at least, you know, until the, the canonical book about it was written and then he exposed all our secrets, but, there are different methods that are more reliable than others.

But I think getting rid of that ego is definitely the biggest part for any trader, you know, quantitative or not. You know, as you were mentioning, you get lucky at the beginning. And that gives you all sorts of bad biases and behaviors. And being lucky from the start, may be the worst thing that happens to you over the long-term.

[01:19:46]Rodrigo Gordillo: Yeah.

Yeah. I think Mr. Butler experienced that in ‘99. The king of the trading floor. Well, if you want to hear his story

[01:19:55]Adam Butler: Right. Carrying me around with my inflatable Burger King crown on.

[01:20:02]Rodrigo Gordillo: Yeah, there’s a, if you go to our podcast and look up,

I think, I think Corey had a great, uh, Corey Hoffstein and in Flirting With Models podcast, the first interview you did with you, you go into detail on that one. That’s if anybody wants to hear the backstory, that’s a great one there. Um, Yeah. I mean, this is…

[01:20:18]Adam Butler: Oh, I have lots of backstories about keeping me humble. That’s, uh, those are not hard to find.

[01:20:23]Rodrigo Gordillo: hey, I started, I started in the…

[01:20:25]Michael Philbrick: I’ve got a few of those stories too. If it…

[01:20:27]Rodrigo Gordillo: …stories…

[01:20:29]Michael Philbrick: Both about Adam and myself. So…

[01:20:32]Adam Butler: Yes,

[01:20:34]Michael Philbrick: Well guys, we’re at about an hour and 20 minutes too.

[01:20:36]Michael Robbins: Oh, wow.

That went …

[01:20:38]Michael Philbrick: …wrapping up. We have a lot of fun, Michael. But I want to know, like, so you’ve got this body of work. We’ve covered a lot of topics. We got, let’s take a few minutes and just what do we, what have we missed?

What are the things that you’ve been asked or what are the things that have really kind of created gravity around what you’ve published that we may not have touched on and maybe we’ll rely on your expertise to round it out a bit. What have we missed in this discussion with you? And what would you like to highlight on top of what we’ve talked about today?

[01:21:06]Michael Robbins: I think you were really great about touching on a lot of important topics. So more to not expose things that you missed, but to reemphasize some of the things you said, that the world is a lot bigger than most people give it credit for. And we do have a systematic bias towards thinking we know more than we know.

And what I tried to do in the book was to try to open people’s eyes to all those things. And the biggest criticisms I get about the book is it doesn’t say enough about what I wanted to learn. Well, that’s exactly what I was trying not to do, was to teach you the things that you don’t want to learn, because there’s so much out there that is expensive to learn by experience and people just have to learn it every generation.

They have to re-learn what the previous generations learn and cost them and you see it happening in crypto, right? They’re just reinventing traditional finance the hard way.

[01:22:05]Rodrigo Gordillo: Yeah, and so many of that, you know, the huddle environment is the overconfidence of a trade that worked really, really well. And they’re writing it all the way down or have, a lot of them have, some of them got out. Yeah, I mean, this is a book that I think everybody that cares to learn what’s coming up next in the next generation of investing needs to read, whether you’re planning on doing it yourself or actually overseeing or possibly allocating to quant. You need to at least have a cursory understanding of how to think about the problem. So I highly recommend everybody read Michael’s book. Now Michael, in the last five minutes I do want you to list out every alpha generating strategy and arbitrage opportunity that’s on your desk right now in extreme detail.

[01:22:58]Michael Robbins: You’ve seen that paper. It’s like, what, 105 trading strategies or something on SSRN, where they give an equation for each one.

[01:23:07]Rodrigo Gordillo: I have not seen…

[01:23:08]Michael Robbins: …a funny.

[01:23:09]Rodrigo Gordillo: I’m sure it’s a gold…

[01:23:10]Michael Robbins: right up there. Yeah, that that’s almost as good as, the one about the factor zoo. That has like 30,000 factors in it.

[01:23:22]Rodrigo Gordillo: Yeah.

[01:23:24]Michael Philbrick: I’m also reminded of that quote from The Art of Motorcycle Maintenance, where the guy says, we take a handful of sand from an endless landscape of awareness around us…

[01:23:36]Rodrigo Gordillo: …at our

[01:23:37]Michael Philbrick: …that handful of sand, the world, right? You  think, you know the world you’re holding it in your hand. But you’re standing on a beach. And I think that’s, that is kind of what you know, in a nutshell, what you’re saying. And I love that quote. Cause I feel that all the time. I think I know something. I think as we age, I get, I’m maybe skewing the other way. We’re like, I don’t know anything. I’ve, I’ve lost all, you know, I, yeah, sure. Everything could be true. Everything could be false. So something you have to kind of wrestle with as you go through. But I really love that quote because we, we think we have it all in our hand and we’re staying on the beach…

[01:24:15]Rodrigo Gordillo: Now let’s anchor back to some reality, which is, can we just go back? Yes, there is some reality. You need some bonds for bull markets and some equities for bull markets, some bonds for non-inflationary bear markets, and some alternatives for everything else. And just be thoughtful about your allocations and, and that’s it for now.

That’s a good high level anchor. There is some reality and some things that we can count on. Then if you want to get into the nitty gritty of all the arbitrage opportunities, then you can get lost in the handful of sand, I guess.

[01:24:50]Michael Robbins: A rabbit hole.

[01:24:51]Michael Philbrick: Yeah. Well, that’s awesome. Michael Robbins, guru. Thank you for coming today, spending this time with us. Really appreciate it. Love the conversation and make sure everybody, Like and Subscribe and Share this with others so that we can continue to get great guests.

[01:25:12]Rodrigo Gordillo: Right.

[01:25:13]Michael Robbins: Thanks so much for having me. It was a lot of fun.

[01:25:15]Adam Butler: Thanks, Michael. Thanks, guys.

[01:25:17]Rodrigo Gordillo: Thanks guys.

[01:25:18]Michael Philbrick: All right.

Show more

*ReSolve Global refers to ReSolve Asset Management SEZC (Cayman) which is registered with the Commodity Futures Trading Commission as a commodity trading advisor and commodity pool operator. This registration is administered through the National Futures Association (“NFA”). Further, ReSolve Global is a registered person with the Cayman Islands Monetary Authority.