Big Data, Machine Learning, and AI in Portfolio Management

Ben Johnson: Hi. I'm Ben Johnson, director of global ETF research with Morningstar. Today I'm on the sidelines of Morningstar's 30th annual Morningstar Investment Conference, and I'm joined by Kevin Franklin. Kevin is a portfolio manager with BlackRock.

Kevin, thanks for being here today.

Kevin Franklin: Great to be here.

Johnson: Kevin, our conversation today focused on some topics that are very popular right now in the industry and in the press at large. Those being big data, machine learning, and artificial intelligence. Can you help me understand what are these things? How do you define them? Are they anything new?

Franklin: Absolutely. Great question. I think those three words I think really hit on two big trends that are impacting every aspect of our lives today. I think of big data really as just a recognition that the amount of data in our lives is growing at exponential pace, and you see that in every aspect of your life.

But in investing, it's meaning that investors can't just focus on fundamentals or price data. They need to understand what's happening with ETFs. They need to understand online activity. What are people in social media saying about the security? They need to understand the digital exhaust that we're all exhibiting in terms of when we go into a store. Big data is really the fact data has just gotten massive and more unstructured in nature.

Then you go to machine learning, which I think is a very concrete concept that has been around a long time. It relates to a process that updates itself, a statistical process that just learns when it's given more data. We've been really using those type of techniques for decades, but the truth is that the techniques are getting better and better, and the computing power that you can deploy is also cheaper and cheaper and very easy to apply with cloud-based compute.

Those are the first two terms. I think artificial intelligence is, in many ways, the most interesting of the three because, really, its definition has evolved over time. The textbook definition of AI is when a machine process achieves human level ability at something that we thought humans should do. If you think about the arc of history and I'll focus on finance, 30 years ago, you thought that, really, a human should put together a portfolio. Today, we all probably have some exposure to indexed portfolios that are programmatically built. We wouldn't think of that as AI today. That doesn't sound right.

Really, I think when people use the term AI, they're referring more specifically to strong AI, which relates to the point at which a machine exhibits general intelligence like a human. I think when we look at the AI today, most of it is about doing a phenomenal job of fitting data, which is kind of what these tools have done all along. AI of the future, I believe, will start asking more interesting questions like why is the data like that; will exhibit the ability to be creative; will ask will it have an imagination. That's what people think of when they talk about AI. We don't think we're there yet, but we do, as a firm, really believe that advances in AI are going to be critical for investing.

Earlier this year, we've had a partnership with a few professors at Stanford for five or six years now. We formalized that. We opened an office in Palo Alto, the BlackRock Artificial Intelligence Lab, to really take our efforts in the AI space to the next level.

Johnson: While I think to many all these concepts are very exciting and shiny and new, you're applying these tools in the confines or the construct of active portfolio management, using this for purposes of security selection, for portfolio construction. How do investors understand whether or not you're applying them effectively? I would think, at the end of the day, that the arithmetic of active management still applies here and that there will be some that apply these skillfully and wind up beating the market; there are others that might not, and there will inevitably be cycles in between. So, how do investors get comfortable with the efficacy as measured by market-beating returns, you name it, of your application, of all of these various tools, when it comes to managing money?

Franklin: That's another great question. I always go back to the client proposition. What are we trying to deliver to our clients? In my team, it's very clear. There are two goals. One, we want to deliver consistent outperformance. We want to beat that market index as frequently as we can. The second, we want that out performance to be different from what you're getting from our competitors or from an average active manager.

I think the proof is in the pudding. We assess our own performance and we ask the question, how consistent have we delivered alpha? And is it different from what you get from the basic factors, whether it's low volatility, generic value, and small cap strategies?" When we look at our 30-plus-year history of generating alpha for institutional clients and now moving into the retail space, we have, in the last five to seven years, delivered the most consistent performance that is also the most different from what you're able to get from our average competitor. Our 10-plus-year effort really investing in this space has paid off and our clients are seeing it.

Johnson: Kevin, I want to thank you again for joining us today. It's been a terrific conversation.

Franklin: Thank you.

Johnson: For Morningstar, I'm Ben Johnson.