Quantitative analysis, more than any other sector in finance, is a numbers game. To be a successful quant means implementing a cutting-edge portfolio strategy backed by solid data. And with large industry players investing billions of dollars into AI in recent months, it's becoming more critical than ever for early and mid-career quants to embrace AI-human collaboration or risk getting left behind.
Columbia Business School's Kent Daniel knows quant inside and out. As the Jean-Marie Eveillard/ First Eagle Investment Management Professor of Business, Daniel has taught courses in behavioral finance and capital markets and investments.
Before joining CBS, Daniel worked with the quantitative investment strategies group at Goldman Sachs Asset Management. He became a managing director and head of the firm's QIS equity research effort in 2005 and a co-chief investment officer in 2009. He has also served as an associate editor for the Journal of Finance and director of the American Finance Association and the Western Finance Association.
To understand AI's current and future impact on the quant field, and how quants can stay ahead of the curve, Columbia Business spoke with Daniel, who is currently a research associate at the National Bureau of Economic Research.
“ An AI-based strategy that is implemented poorly and that doesn't use good data isn't going to perform well. On the other hand, to say, 'I'm going to ignore AI. I'm not going to use it. I'm not going to pay any attention. I don't want to use all of the data out there'— that's also a very bad approach.” - Kent Daniel, the Jean-Marie Eveillard/First Eagle Investment Management Professor of Business
Professor Kent Daniel
CBS: A lot of business students, and those who are in the early stages of their careers, are wondering if AI is going to take their jobs. Is that something that could happen to quants?
Kent Daniel: My sense is no. I think AI will definitely take some jobs. If you're doing a routine job that requires no thought, for example, that can basically be automated. And if you're doing the equivalent of working in an assembly line in a factory, then those are the kinds of jobs that could be replaced by AI-based code. That is not going to be the case for the students that CBS produces.
You might read a lot about the idea that AI is going to be better than humans and that it will come to dominate humanity. When it comes to the implementation of a portfolio strategy, that's just a misperception of the way things work. Any AI-based system is going to be successful only to the extent that there are no better implementations around. It's going to be the case that if you can come along and design a better implementation, you're going to do better. Because anything that's inferior in a free market is going to be competed away. And I know that even when I was working in quant, people had the idea, “Oh, quant is the solution to everything. If you can do quant, you can rule the world,” and that's just wrong.
If you have a bad quantitative implementation, you are going to lose to other implementations. It's all about the inputs, and there's a saying, “garbage in, garbage out.” That's true for quantitative portfolio strategies, and it's true for AI strategies as well. So how precisely you train any AI-based system—what data you're using and how specifically you use that data—is going to determine whether it's successful or not.
It's even more pronounced in quantitative portfolio implementations because there's this idea that we've got so much data, and we'll design the AI system to go and look at the data and figure out what the best strategy is. Well, this is something that just doesn't work in financial markets because in reality, we have very little data. The reason is that the world is constantly changing. When you're designing a portfolio strategy, you're not competing against a static world where everybody continues to do the same thing. Everybody else is trying to implement and innovate new strategies.
CBS: So it's important to consider the externalities?
Daniel: Exactly. Unless you implement a strategy that is better than what everybody else is doing, you're not going to have particularly good performance. I think what our students will need to be successful is to understand all of these arguments, to understand the quantitative techniques that you need to use. If you don't know how to do that, you're going to be in trouble.
At some firms, you might hear, “Well, I'll just hire people who are good at using big data and that's how I'll deal with it.” The problem is there are lots of people who have misperceptions about the way that you need to use data. You don't really understand that if you use big data the wrong way, you can convince yourself to do a lot of things that are wrong. And so really understanding the ins and outs of how to use data, make informed decisions, and use good judgment is incredibly important. And I think we're already [teaching that]—we've got a bunch of really good coursework. We do a good job training our students. And our curriculum is evolving continuously, so our students can continue to do really well on this dimension going forward.
Professor Kent Daniel
CBS: Do you believe it is more about the quality of the data and how you use it or how one uses AI in general?
Daniel: An AI-based strategy that is implemented poorly and that doesn't use good data isn't going to perform well. On the other hand, to say, “I'm going to ignore AI. I'm not going to use it. I'm not going to pay any attention. I don't want to use all of the data out there”—that's also a very bad approach.
You need to be a really intelligent consumer and a really intelligent user of data and AI implementation techniques to be successful.
CBS: Is the introduction of AI into quant going to democratize the industry, or is it benefiting the larger players more?
Daniel: It will definitely change the industry. I think you may see some new entrants who are doing things in a different and better way, and if the bigger players don't respond, then they could very easily lose any edge that they already have. So, in that sense, it could be a democratizing influence.
On the other hand, it could be that there are returns to scale here. So, to the extent that you can build a really good group who really understands this field, you can leverage those advantages and I think be super-successful. There are some good economies of scale, so it could go either way. One thing is for sure: If you're not paying attention to this while building a quant shop, you need to.
“You might read a lot about the idea that AI is going to be better than humans and that it will come to dominate humanity. When it comes to the implementation of a portfolio strategy, that's just a misperception of the way things work.” - Kent Daniel, the Jean-Marie Eveillard/First Eagle Investment Management Professor of Business
CBS: Does AI open the door for quants to potentially neglect softer fundamentals such as the quality of a firm's management?
Daniel: There's an old joke about quants: You see someone looking under a streetlight at night and you ask, “What are you doing?” And they say, “Oh, I'm looking for my keys.” And you ask, “Well, do you think you lost them under this streetlamp?” And the person says, “No, I lost them a couple blocks back, but there's no light there.”
Historically, I think there's a little bit of a tendency to say, “What data do I have? How do I use that to make good decisions?” Not enough time is spent on thinking about how you can process “soft” data on management quality and things like that.
In my class, I will be focusing on how you deal with soft data. We'll take some of the top value-investing techniques that CBS is renowned for and ask how we can make them a part of a quant process. A good quant process should be able to incorporate all different kinds of data, but if they're tricky to quantify, people have often left them out. And I think management quality is a really good example of something that's important and can give you a moat, or a barrier to entry. If you can do a better job thinking about that than somebody else, then you are going to earn better returns.
CBS: What are the ethical issues in integrating AI into quant? Are there biases that can appear in the data?
Daniel: Standard quantitative processes use statistical techniques to identify correlations. A good example would be the admissions process for a university or a business school, where if you just rely on correlations and don't go in and process the data in a more thoughtful, deeper way, you are going to end up with lots of unintended biases. Part of what I want to emphasize in this course is that if you go and implement a statistically-based algorithm and don't think carefully about where these correlations come from, you're going to miss a lot of very valuable opportunities.
I spent a few years working in the quant industry, and it's as much about what can go wrong in quant as it is how you do a quant process. A bias means that you are making decisions on the basis of some variable that is maybe correlated with something else but that doesn't really reveal the underlying truth about some situations. A really well-thought-out quantitative process should use all the data you can gather in the best possible way. I'm more optimistic that a process like that is going to avoid these unintended biases.
I think the way to do that is to think carefully about the way the quantitative process you have works, how it's going to process the data, and what the outputs are going to look like. It's just like any sort of regulatory work, any sort of legal issue: If you just rely on data that's very broad, and you don't dig down into the details with the goal of understanding the real relationships, you are going to introduce biases.
CBS: What advice do you have for those who are more established in their careers? How is AI going to affect what they do?
Daniel: You want to always do what's best for your customer—that's what makes a successful business. The businesses that look at every innovation and think about how they can use it to make their customers better off, to make better investment decisions, or to manage their financial future better are the ones that are going to succeed.
