The Future of AI in Stock Picking

A recent article in Forbes tells the story of EquBot, a San Francisco based ETF advisory firm that uses proprietary software for stock picking.

The firm’s chief executive, Chidananda Katua, got the idea during a business school lecture on hedge funds he attended at UC Berkeley four years ago: “Khatua imagined that something powerful might come out of the ability to blend precise financial data with the fuzzier information to be found in annual reports and news articles.” He started EquBot with two classmates, $735,000 from angel investors and a $120,000 credit from IBM toward software and hardware purchases.

The EquBot system reportedly “swallows 1.3 million texts a day: news, blogs, social media, SEC filings. IBM’s Watson system digests the language, picking up facts to feed into a knowledge graph of a million nodes.” The system undergoes a trial-and-error process akin to the neural connections in a brain, weighting the importance of different data points. The article explains, “Thus does the system grope its way toward knowing which rippled in input data are felt a week, a month, or a year later, in stock prices,” which could mean as many as half a quadrillion calculations in a day.

EquBot’s software has picked both winners and losers since its launch, the article reports, adding that it’s too early to tell whether the venture will succeed (it manages $120 million). The firm’s U.S. fund (AI Powered Equity ETF) has lagged the S&P 500 by an annualized 3 percentage points, while its international fund (AI Powered International Equity) is 6 points ahead of its index.

While the firm claims that its funds are the only actively managed ETFs using AI, the article concludes that it “won’t have this turf to itself for long. IBM is selling AI up and down Wall Street.”