Our industry has always been quick to recognize a rolling bandwagon, and quicker still to jump aboard. Inevitably, many managers have latched onto big data and AI as a way to shelter under the quant tent. It makes them seem modern, differentiates them from slower adopters, and, most importantly, restores some faded luster to active management. Unfortunately, real understanding of these technologies in our industry is still scarce, and their applications to investing are anything but straightforward.
Our view is that big data is definitely valuable as a new data source but is clearly not a panacea. The problem, as it relates to AI and its nearly synonymous cousin, machine learning, is that these are data mining machines. Stock markets are not perfect information settings. AI is fundamentally about pattern recognition, but the basic truth is that for the great majority of the patterns it finds in the market, there is no cause and effect and therefore no underlying predictive rationale.
Big data and AI are, rather, tools that can be helpful in the investment process when they are used properly. We think the key is having a clear investment philosophy that guides what data sources a manager is using, what new signals are tested, how they are tested and how they fit in the manager’s models to generate meaningful, sustained net outperformance.