Paper discusses alternative data mining methods that can complement traditional style factors
Given the fast-paced advances in computing power and analytics in recent decades, investment professionals have the opportunity to access a richer trove of information to inform their strategies and decision-making than ever before.
That’s a key point driven a new white paper from MSCI, written by Peter Zangari, global head of Research and Product Development.
In the paper titled Technology and Generational Change for Investors, Zangari highlighted how “the acceleration of data collection, and machines’ ability to process it” has opened the possibility for investors to make decisions based on an ever-growing number of inputs.
“Even the algorithms used, which have been around for decades and seemingly wouldn’t change, may evolve as well,” he said.
With the explosion of alternative data and methods to collect it, investors can find themselves overwhelmed and unable to sort between signals and noise. But according to Zangari, the MSCI Research Team has identified three approaches that stood out for “having explanatory power beyond traditional factors.”
While poring through corporate filings for information isn’t new, he noted that natural language processing and machine learning now makes it possible to squeeze more valuable information from machine reading of companies’ 10-k filings or earnings call transcripts. “Depending on an investor’s views or thesis, it’s possible to scan for key words, such as sustainability or diversity,” he said. Paying attention to the frequency of certain words or word patterns, he added, can be used to get a sense of how a company defines itself.
Another possible application is to derive metrics of consumer sentiment toward a company from online mentions. Monitoring reviews, social media, and other content for citations of a company’s products and brands, Zangari said, can potentially allow investors to find correlations between that data and the company’s fundamentals and stock performance.
Finally, there’s the potential of looking at stock transactions by company insiders, including those by key company executives and those that may relate to employee compensation. “We found that this data stood out with unique characteristics when compared to traditional risk model style factors,” Zangari said.