Some investing tasks are easier to automate than others. Anything that can be rules-based (index construction, credit scoring, or standard valuation metrics) is already used by most quantitative investors. Turning qualitative information, such as investor sentiment, management confidence, brand identity and corporate governance, into quantified data is more difficult. My firm sees promise here, through machine learning and specialized forms of analysis. There are possibly some areas, such as forensic accounting and activist investing, where human expertise is too specialized and sample sizes too small for Big Data analytics to be of much use.
As a quantitative investment researcher, I strongly believe in objectivity, empiricism and using the maximum amount of relevant information. Big Data is exciting. In many ways it captures the hopes we have always had about being able to explain the world. There is potential in the information that is becoming available – satellite and location data, text in all forms, social media and crowdsourcing. There will be more ways to gain insight from machine learning, pattern recognition, network analysis and other techniques. But let’s be clear on what it can really do for us in its current form, with our current technologies. The impact on actual investment results is likely to be modest for the near future.
When evaluating investment approaches and strategies, your best bet is to look first at the human resources behind them. I suggest we investors keep our focus where it has always been: on finding smart, insightful people with creative minds and strong analytical skills. One day soon, Big Data will be unremarkably widespread and easier to use, but not before we make substantial investments in collecting, curating, and thinking about the story behind it.