Goldman Sachs: How Big Data, AI Benefit Investors

Ben Johnson: Hi, I'm Ben Johnson, director of global ETF research with Morningstar. I'm here on the sidelines of the 30th annual Morningstar Investment Conference. I'm joined by Osman Ali. Osman is a portfolio manager with Goldman Sachs.

Osman, thank you for joining me.

Osman Ali: Thank you for having us, Ben.

Johnson: Earlier today we were talking about big data, machine learning, and artificial intelligence, these big, scary, looming, lofty concepts that have been dominating the headlines in the popular press. People are worried about being replaced by machines, this age-old battle between man and machine. But you made some important points specifically about the tug-of-war between man and machine and that we as flesh and blood human beings probably shouldn't too terribly worried about the fates of our profession, and if anything, maybe these tools are something that you can leverage to build better portfolios. Can you talk about the balance between man and machine, how you and your team make use of all of these tools when it comes to building portfolios?

Ali: I can speak for what we do within the context of investment management, perhaps not so generally across all industries, because I'm sure innovation and enhancements that we are making across technology will have disruptive effects in different parts. But within our investment team, we've always been of the belief that it takes an experienced investor or a portfolio manager with expertise in markets to be able to get the most out of big data, alternative data, to be able to make the best decisions with respect to how to use machine learning and AI.

In fact, we've built an investment culture on hiring and empowering portfolio managers to go look for new data sources to use, to go look for techniques that can better harvest that. Our team, I think, it's safe to say, is larger now in terms of individual researchers and portfolio managers than it was five years ago. Not only that, it's expanded to new skill sets. We have data scientists which we didn't have 10 years ago that are a part of our investment team. We have a larger technology team than we ever did.

We have, kind of, confronted this new world of data, technology, and AI by perhaps changing some of the skill sets with which we've hired. But growing our team so we have the right people in the right positions to be able to take the best advantage of these new enhancements.

Johnson: When you think about the investments that your firm has made in people and in tools, you are not alone; you have a number of different peers, a number of different competitors that are making similar investments. There's been talk about this turning into a bit of an arms race of sorts. What are your thoughts on whether or not this is an arms race and whether if at the end of this road as everyone tries to look at the same data sets, leverage the same tools, develop the same sort of incremental additions to their human capital, that the net effect is sort of a crowding, a washing out of any sort of alpha that might be derived from pulling of all of this together in building portfolios?

Ali: That's a very good question. There's two things that come to mind. One is the framing of how much capital there may actually be that is taking advantage of all these alternative data sets. The second is, what have we observed of the erosion perhaps of the efficacy of some of these alternatives data sets.

On the first point, if you look at the entire broader marketplace and you divide that between active and passive, you are going to find that passive is very significant if not overwhelmingly large part of the market. Then you've got a set of active managers that are out there that are trying to capitalize on informational inefficiencies and risk premia. Of those active managers there may be a lot of active managers that are perhaps talking about big data and AI, but if you really look at it, there's a much smaller set that can offer any sort of convincing proof that they are in the right position to do so. By convincing proof I mean a data and technology platform that enables that kinds of enhancements. It's not cheap. It requires certain skill sets. Not everyone can do it. Then you end up with a pretty small group of investors that can truly articulate their advantage in the big data and the AI realm which is a very, very small portion of the total investment public out there.

Nonetheless, I think, you should be very cautious of anything that smells that crowding. We certainly are. We monitor how much some of these alternative data sets and the signals that one can derive from them are in fact crowded, how much value-add did we get from credit card transaction data a few years ago versus what we are getting today, satellite imagery data a few years ago versus we get today. We are actually finding that the erosion of that data's ability to predict returns is relatively low. We are not seeing a lot of crowding. A big part of that is the people that are using that data are all using it in very different ways. There's no real rule book or playbook for how you take satellite imagery data and convert it into an investment signal as it was perhaps with more common factors that existed 10 years ago.

While people maybe talking about these data sets, everyone's implementation is different. I think the good thing for clients around the world is that there's different ways of making money with the exact same data sets. While we are paranoid about crowding and are looking out for it, we are not seeing it quite yet in some of the advanced alternative data sets that we are taking advantage of in our business.

Johnson: Well, that's a fantastic insight. We appreciate you being here to share your insights with us. We'd love to have you back in the future.

Ali: Than you, Ben, for having us.

Johnson: For Morningstar, I'm Ben Johnson.