Chris Marx Hi, I’m Chris Marx, senior investment strategist at AllianceBernstein, and I have the pleasure today of having Jonathan Berkow with me to talk about our efforts in data science.
It seems like there’s almost an intimidating amount of data out there. How do you go about trying to prioritize what you’d want to get it and separate where you can gain real insight from what might just be noise around us?
Jonathan Berkow What we try to do is something we call question-driven data science. it’s important for our data scientists to work very closely with the fundamental analysts, identify what the question is to identify what the key controversies are, and then help them find the data and the tools to help answer that question.
Chris Marx So it sounds almost like the question is as important as the data and has to almost come first in the process.
Jonathan Berkow That’s exactly right. And recently, we’ve gotten lots of interesting questions from our analysts. Some are very broad and some are a little bit more narrow. On the broad side, we have analysts who are curious about the global activity. How is the activity level of the world changing? We’ve also gotten questions that are very narrow, so for example, one analyst who is covering banks, wants to understand the exposure of its banks to colleges. And so, the risk is that colleges may or may not reopen in the fall. So, if a bank has lots of exposure to their deposit base in those towns that are college towns, that may be a huge risk for that particular security. And so what we did was to use geospatial information about the locations of the banks, locations of the colleges, size of the college and the size of the popular, local population to determine which of those banks and colleges were at major risk factors. So, our analysts were able to process all of the information that would’ve taken them weeks and weeks to do on their own, very quickly.
Chris Marx Right. I think back to some of the tasks I had as an analyst, and it certainly seems like it makes it much more efficient to do things that would have taken us a long period of time. You can compress that and answer the same question even more quickly or with more data. It’s fascinating.
So, I know one of the projects that you worked on was regarding the auto industry. Maybe you can tell us a little bit about the work you did there.
Jonathan Berkow Yes. This was one that we used interesting alternative data from around the globe to help us have better confidence in our views about American auto parts retailers. What we noticed in China is that their car congestion activity monitors were recovering quite quickly. But public transit was something that was lagging. In looking at the US data, we actually saw similar trends. And so, what this tells us is that going forward over the next few years, there’s probably going to be an increase in demand in, for car parts and auto repairs because people are moving away from subways. They’re basically swapping subways for cars. And so that gave us a lot more confidence in our view that that could be an area of growth for our positions.
Chris Marx So, Jonathan, does this work end up impacting actual portfolio decisions?
Jonathan Berkow Yes, absolutely. We want to make sure all of our analysis is useful in investment decisions. A recent example is that our analysts were interested in a German real estate investment company that had holdings in German hotel chains. And so, we were able to identify is that the search interest for these hotel chains had already bottomed and was starting to recover. And so, this was something that gave us confidence that that hotel chain would benefit over the next three to six months.
Chris Marx Really fascinating stuff. I just want to thank you so much for your time.
Jonathan Berkow Thank you very much.