Finding big alpha in big data

01 oct 2015
por Bill MacCartney, PhD, Bradley J. Betts, Jeff Shen, PhD, Raffaele Savi

The ‘big data’ concept

The world is awash in data. We are generating some 2.5 quintillion (2.5 x 1018) bytes of it every day, IBM estimates that about 90% of the data in the world today has been created in the last two years. So it is no surprise that it seems that every day generates some new promise of how we can use this ‘big data’ to change the world.

Investment implications

What really matters to us from an investment perspective is how data can be put to work to generate alpha. In practice we are interested in all kinds of data, particularly as they relate to human economic behaviour. Given the size of the datasets being created, it is necessary to use new tools and research methodologies.

These tools can help us to both understand what human indicators, such as analysts and employees, are trying to tell us about individual companies, as well as helping us construct better economic indicators that may offer clues for the outlook of entire industries.

Having superior data and superior methods of extracting information from that data have provided an advantage in investing, and we believe they always will. This is not unique to the class of signals that draw on machine learning, natural language processing. We will use these tools for as long as they give us alpha. If they stop doing so, we will use other tools. That’s the nature of good science. Our allegiance is to finding alpha, not a particular set of tools or ideas.

Sustainable alpha

We firmly believe that a data-driven, scientifically based and technologically aware research culture can produce sustainable alpha. Furthermore, we are just beginning to scratch the surface with regard to understanding everything that both data and humans can tell us about markets. We also expect that as investment managers understand more about how the power of artificial intelligence and machine learning and conditional thinking can be used, important new insights will be revealed.