Stay up thus far with free updates
Simply register for Fund management myFT Digest – delivered straight to your inbox.
Asset managers are increasingly using artificial intelligence to guide investment decisions, track portfolio managers' habits and discover money-making opportunities.
JPMorgan plans to expand its use of a generative AI tool later this 12 months to flag questionable decisions by portfolio managers, resembling potentially selling top-performing stocks too early, company officials told the Financial Times.
The tool, called “Moneyball,” is designed to indicate portfolio managers “how they and the market have behaved in similar situations and help them correct distortions and improve their processes,” said Kristian West, head of the investment platform at JPMorgan Asset Management.
Other fund managers are using AI to enhance human analysts, discover targets for litigation funding and explain allocation decisions to investors.
These diverse efforts display how the AI war in asset management is shifting from paper-intensive compliance and marketing tasks to supporting investment professionals in decision-making.
JPMorgan's tool is a pilot program that continues to be under development and shall be made available to a wider range of portfolio managers later this 12 months. It is a component of the $3.2 trillion asset manager's Spectrum portfolio management platform, which relies on around 40 years of knowledge.
Voya Investment Management has already implemented a virtual analyst that may monitor stocks for potential risks, complementing the $331 billion manager's human research staff. Portfolio managers have access to a dashboard that may display a human analyst's review of securities together with AI feedback, resembling when a stock is deemed dangerous.
So far, Voya's AI analyst has shown an excellent ratio of right and incorrect decisions, making its alerts “a high-quality signal,” said Gareth Shepherd, co-head of machine intelligence at Voya. He compared the method to a pilot and co-pilot reading signals from an airplane's flight management system.
“The flight management system supports the pilot in his decisions, however the pilot has the ultimate say,” said Shepherd.
Legalist runs a $1 billion hedge fund focused on litigation financing. The company uses a proprietary AI search tool called the Truffle Sniffer to seek out attractive investment targets amid a flood of civil lawsuits.
The “sniffer” searches court records for signs of a good consequence, resembling friendly judges and trial classes, but in addition pretrial decisions that indicate particularly strong cases.
“We search for cases where there are clear signs that the plaintiffs are winning but the cash has not yet been collected,” Eva Shang, co-founder of Legalist, told the FT.
In some cases, AI has an enormous say, resembling an AI-powered exchange-traded fund from South Korean conglomerate LG and SoftBank-backed Qraft Technologies.
Their LQAI ETF, launched in November and based on a proprietary AI stock-picking tool, has been further developed and now includes an AI-generated monthly holdings report. A recent AI-generated report explains the fund's decisions to favor certain corporations and sectors and sell others.
“As portfolio manager of LQAI, I actually have increased my investments in resilient and technologically advanced corporations resembling (Google parent company Alphabet) while barely reducing exposure to corporations with traditional media problems. This reflects a cautious but optimistic approach to capitalizing on growth opportunities amid financial volatility,” the AI-generated holdings report said.
Despite these developments, there are also skeptics about AI's potential to extend long-term returns for asset managers.
Veteran portfolio manager David Giroux, who runs the $59 billion T Rowe Price Capital Appreciation fund, argues that almost all of the AI-focused mental capital in asset management is concentrated on achieving short-term competitive advantage, quite than the harder task of estimating earnings potential years into the longer term.
“I imagine AI can do very, little or no to deal with this inefficiency,” Giroux said.
