How elite investors use artificial intelligence and machine learning to gain an edge

Citigroup (C) uses machine learning to make portfolio recommendations to clients. High-frequency trading firms rely on machine learning tools to rapidly read and react to financial markets. And quant shops like PanAgora Asset Management have developed complex algorithms to test sophisticated investment ideas.

Much of the technology that elite investors use isn’t really new. Financial firms are just better able to harness the power of AI and machine learning because today’s computers can process information much faster. And there now exists vastly more data than there did years ago.

The rise of machine learning

Still, technology is rapidly disrupting the financial industry — and will continue to do so.

“The rise of machine learning will really make our industry unrecognizable in the future,” said Anthony Cowell, head of asset management for KPMG in the Cayman Islands. His clients include some of the world’s largest asset managers, hedge funds and private-equity firms.

For instance, Citi Private Bank has deployed machine learning to help financial advisors answer a question they’re frequently asked: What are other investors doing with their money? By using technology, the bank can anonymously share portfolio moves being made by clients all over the planet.

“Traditionally that kind of information was sourced from your network. You might have had a few coffees or heard about it over a cocktail,” Philip Watson, head of the global investment lab at Citi and chief innovation officer at Citi Private Bank, told CNN Business. “Now, we can share insight that is very valuable.”

Citi also built a recommender engine that uses machine learning tools to advise clients. The platform recommends tailored research reports, solutions and even alerts clients of major events such as the maturity of a bond in their portfolio.

Machines assist high-speed traders

Domeyard, a Boston hedge fund that focuses on high-frequency trading, depends on machine learning to decipher 300 million data points in the New York Stock Exchange’s opening hour of trading alone.

“We rely on the help of machines to make easier and faster predictions of what will happen in the next second or minute,” said Christine Qi, Domeyard’s co-founder and partner.

But Qi cautioned that machines are “only as smart as the data you’re feeding it.”

Earlier this year, PanAgora, the Boston quant shop, expanded its exposure to China by launching a “self-learning” algorithm that deciphers Chinese “cyber slang” used by investors on social media to get around government censorship, Chen said. The findings give portfolio managers at PanAgora a valuable window into sentiment among retail investors, who dominate the market in China.

Man vs. machine?

Technology executives warn not to believe all the hype about artificial intelligence and machine learning — especially about robots taking over.

“Some of the effects can be wildly exaggerated,” said Citi’s Watson. “It’s a human plus machine world. It’s not a machine-only model. Nor do I see it becoming a machine-only model for a long, long time.”

PanAgora’s Chen agreed. “It’s not man versus machine. It’s man plus machine.”

At PanAgora, humans have the final say on investment decisions and at times override what the computer models tell them to do.

“Machines are not sentient. Terminators are not going to rise up and kill us all in the next 10 years,” Chen said. “I hope.”

Most jobs will be impacted

But that doesn’t mean humans won’t be disrupted.

Foster said that the most optimistic outcome is that businesses, governments and education systems get ahead of this disruption by re-skilling workers.

“Probably the world is moving more slowly than that. There is a risk that people will be left behind,” Foster said. “It’s incumbent upon us in business that we’re helping our workforces get ahead of the curve.”

Rather than getting outright displaced, Citi’s Watson thinks many workers doing menial back-office jobs could be moved to more rewarding positions.

What’s next?

“We actually don’t know yet,” Jackson said when asked for specific use cases. “We’re just beginning to understand the power of this.”

He said it’s already clear that quantum computers will excel in several areas: encryption, security, chemistry and machine learning.

“It will live up to the hype,” Jackson said.

There are still many things that computers can’t do in the financial realm.

For instance, sophisticated investors often use game theory to map out how other market players will react to a given situation. Game theory allows firms to cash in by positioning themselves — before sharp market swings occur.

PanAgora’s Chen said that machines can’t do that — yet.

“I hope to see it in the next five to 15 years,” he said.