Point72 Asset Management, billionaire Steve Cohen's family office, recently brought on Matthew Granade to oversee big data.
Granade oversees data ranging from sell-side research to alternative data that is then incorporated into the firm's quant and fundamental investing strategies.
Point72 manages $11.6 billion, the bulk of which is Cohen's fortune.
Business Insider caught up with Granade, the firm's chief market intelligence officer, last week at the investment firm's New York offices. What follows is an excerpt from the conversation, which Business Insider has edited for clarity and length.
Rachael Levy: Walk me through how Point72 views quant investing.
Matthew Granade: Our longest-serving quant portfolio manager has been here 21 years. And the reason I mention that is there has been a very, very long-standing commitment to quantitative-style investing here. We're very known for our discretionary investing, but Steve has been doing quant investing for a very long time. Cubist, which is run by Ross Garon, is a standalone business. It operates separately from the discretionary long-short business, but quant more broadly, increasingly quant techniques and quant ideas, are infusing themselves everywhere.
Cubist is doing very traditional quant, like stat arb [statistical arbitrage] and using traditional quant data sets. This idea of building equations and using machine learning and breaking your process down, you're starting to see that pop up in the discretionary side of the house as well.
But the difference is the discretionary folks take a very company-centric view of the world. So an analyst on the discretionary side will cover 40 or 50 stocks and will know each of those companies really well, whereas a quant trader will trade 2,000 or 3,000 stocks — they tend to know like a data set or a technique, whereas our discretionary investors really, really know the companies.
But what we're increasingly finding is that data and some of the techniques you can apply to the data are very powerful in the discretionary investment process.
Levy: Would you qualify that as "quantamental"?
Granade: I try not to use the quantamental word because it's sort of more obscuring than clarifying sometimes, but yeah, I think that's exactly right. Sometimes, people present themselves as quantamental and they're not really doing anything that different. It's sort of a different name on old-style quant. But I do think there's a new thing forming, which is fundamental company understanding, quant techniques, alternative data all sort of being done by one person or one team. If you want to call that quantamental, that's my definition of quantamental.
It was funny, I was using "quantamental," but I started encountering more and more people using "quantamental" and not really meaning what I meant, and so then I just stopped. So now I encounter people here and they say, "Isn't that what we call quantamental?" Right now, I don't have a good term.
Levy: So quant techniques using alternative data. You mean using algorithms to parse through this alternative data. The quant techniques were there already. So is the only thing that's new the alternative data?
Granade: I'd say the data is new and the other thing that's new is bringing it all together. Let's say if you go to our most traditional long-short discretionary investor. Most of what they did or do is talk to the companies, get some data from them usually through major releases, talk with the sell side, and the major thing they get from them is models, and then they have an analyst that builds their own model. They might do some more conversations and tweak that model.
Whereas now, what you might do if you're trading Chipotle, let's say, you'd look at the credit-card [data]. Well, that requires a tremendous number of machine-learning techniques to make that data useful and a tremendous number of data science and statistical techniques to make it insightful, and then you get to your financial model and you couple that maybe with geolocation data or some other thing. So that's the new thing.
Traditionally you didn't have someone on a long-short fundamental team who had anywhere near the skills to do this. They were Excel experts and that was the end of it. Whereas now, we have a lot of teams with data scientists on them.
Levy: So is it the case that you have people that are all in one or do you have a sidekick that is going to help me parse through this Chipotle data that would take me years to go through?
Granade: It's kind of both. This team out here, they're building products that everyone on the platform can access. The second thing, instead of hiring an analyst, we are hiring a data scientist. In that case, they're accessing the data directly and building their own models. And the third thing you're seeing is more and more portfolio managers train themselves, with our help sometimes, but also sometimes on their own on all the statistical understanding you need to have to be able to appreciate and use this data well.
Levy: Is it is possible to find the all in one package?
Granade: No, I think it's different things. What you see now is you have a data-science-track-type person, and you have a fundamental-analyst-type person who worked in private equity or investment banking and then came here. And then that person got a masters or Ph.D. in statistics and then went to a startup and came here.
If I think of the portfolio managers of the future, for 10 years from now, I think there's a very good chance that what they do is they have some education in computer science or data science, they then have a job like here in Aperio for three years as a data scientist, then they have some more training as a fundamental analyst and then they start transitioning to be a portfolio manager.
In a seven- to 10-year time frame, the portfolio managers at the top of these things are going to have been trained in all the different pieces, but right now we're still in more of a transition period where we have these two tracks.
Levy: So how does it work if you work here? Do you get training?
Granade: We do data training, statistical training, computer-language training. You have to think the level of the person. If someone is a senior analyst, it's not the right thing to train them to be a programmer — but they do need to understand these data sets and the statistical techniques that can be applied to them.
For our younger analysts, like in the academy, we do teach Excel modeling and a programming language, and we teach statistical techniques and how to do a company interview with the CFO. Those are our 22-year-olds. With our 28-, 30-, 32-year-olds, we do slightly more targeted things.
Levy: How much of quant is just a fad?
Granade: I think quant is a great business. It doesn't come overnight. Steve has been building that business for 21 years. Do I think it's a passing fad? I think people are rushing to the thing that is performing better than other things now — but very few people have the staying power to stay in it. It takes a long-term commitment. You can go look up and see how long DE Shaw and Renaissance [Technologies] have been around. They've all been around since the '90s.
SEE ALSO: This is the future of investing, and you probably can't afford it
DON'T MISS: Behind the life and death of a star money manager accused of insider trading
Join the conversation about this story »
NOW WATCH: The legendary economist who predicted the housing crisis says the US will win the trade war