Some quantitative managers are skeptical that big data, artificial intelligence and machine learning technologies will lead to big breakthroughs for asset managers anytime soon. Quants’ caution is noteworthy given the increasing number of money managers that are investing in data science and offering factor-based funds, widely known as smart beta.
Ryan Caldwell, chief investment officer of Chiron Investment Management, which uses fundamental and quantitative research to run its multi-asset portfolios, says asset managers may be overconfident about the ability to use big data techniques to produce brand new insights beyond well-documented factors at this point.
Meanwhile, BlackRock, the world’s largest asset manager with $5.1 trillion of assets, is in the early stages of a big bet that investors will increasingly want low-cost, systematic portfolios. The firm announced in late March that it’s overhauling its active equity group, consolidating some funds that rely on human stock pickers with those run by quantitative, rules-based algorithms, while investing significant resources into data science capabilities to support the initiative. The industry shift might not be easy.
“There’s a lot of noise in the data and people will have to work through that to get a reasonable signal,” said Caldwell, who ran $45 billion for Waddell & Reed, including the Ivy Asset Strategy, before co-founding Chiron. “Just because you can create more inputs doesn’t mean it will be valid.”
While some signals may be valuable to managers that trade fast and have high turnover, there are limitations to consider, as well, according to Caldwell. For example, cameras set up in a retailer’s parking lot may capture changes in traffic that give an investor an information edge in the short term. “But it could be meaningless as to whether a company is able to beat quarterly earnings estimates,” he said.
Quantitative Management Associates, the $116 billion quantitative manager founded in 1975 that is part of PGIM, tests about 10,000 signals before it finds the one it will add to its models, said Joshua Livnat, a managing director at the firm who focuses on global accounting research. There’s been an explosion in data over the past five years, he said, so the firm focuses on samples that have a history, are released frequently, and have breadth. This means the data apply to as many companies as possible, not just a small subset. For example, there’s a lot of talk about using Twitter and other social media information, but there’s no history to that data, Livnat said.
Asset managers are embracing factor investing partly in response to the overwhelming popularity of passive funds and the sustained inability of the average active manager to beat them. Computers can be programmed to search the markets for stocks that exhibit factors, or characteristics such as value or momentum, which academic research has found to be the source of investors’ returns.
In the hyper-competitive world of today’s markets, managers are hoping to find more factors, refine the measures behind them, or discover new predictive patterns in data, through technologies like artificial intelligence, to gain an edge over rivals. Big advances driving outperformance, though, may prove elusive.
