Can an Algorithm Replace Stock Analysts?

From left: Kris Hammond, Larry Birnbaum, and Stuart Frankel of Narrative Science in Chicago, 2011 Photograph by Peter Wynn Thompson/The New York Times via Redux

As a journalist, I naturally regard the Chicago-based Narrative Science with morbid fascination. In 2010 the startup made news selling software that read baseball box scores and generated newspaper articles from the data. Narrative Science says the idea isn’t to put reporters out of work, but to extend their reach to events that would otherwise go uncovered. That didn’t stop Bloomberg Businessweek from asking whether sportswriters were really necessary.

And not just sportswriters. The company’s technology, called Quill, can also render stories from crime stats, medical studies, and survey data. Forbes publishes Narrative Science stories based on public companies’ financial statements. Now the company says it doesn’t just want to do the job of business journalists, but also the work done by financial analysts.

At yesterday’s demo day for the FinTech Innovation Lab, a New York-based program that matches tech startups with mentors from the banking industry, Narrative Science Chief Technology Officer Kris Hammond said the software could turn data into written reports for wealth management firms, institutional investors, and research shops. “The target is to take the best financial adviser or analyst and turn his skill set into something that the machine is doing,” Hammond says.

Are financial firms buying in? Convincing a newspaper to run a computer-generated sports story is one thing, but asking investors and advisers to trust machine-written language seems like a harder sell. Maybe not, says Hammond. One-third of Narrative Science’s 50 clients are in financial services—Hammond declined to name names—and the company is shifting its focus away from media and toward finance. (Narrative Science also raised venture capital recently from In-Q-Tel, which invests on behalf of the Central Intelligence Agency.)

Wall Street already trusts computers to trade millions of dollars worth of securities on their own, so using an algorithm to put the numbers into words might not be a giant leap. Does that mean financial analysts should be as worried about their jobs as sportswriters? “We’ve had, I wouldn’t say zero, but close to zero pushback from analysts afraid of losing their jobs,” Hammond says. “Most of them see it as a way to spend their time working with the high end of their skill set and letting the machine take care of the lower end.”

Before it's here, it's on the Bloomberg Terminal.