Lessons from Finance’s Experience with Artificial Intelligence

It appears that this rule applies to artificial intelligence (AI) and machine learning, which were first employed by hedge funds decades ago, before their recent hype. First came the “quants,” or quantitative investors, who use data and algorithms to pick stocks and make short-term bets on which assets will rise and fall. Two Sigma, a quant fund in New York, is experimenting with these techniques. It was established in 2001. The Man Group, a British outfit with a large quant arm, launched its first machine-learning fund in 2014. AQR Capital Management out of Greenwich, Connecticut started using AI around the same time. Then came the rest of the industry. The hedge fund experience demonstrates AI’s potential to revolutionize business – but also shows that doing so takes time, and that progress can be inhibited.

AI and machine-learning funds seemed like the last step in the march of the robots. The inexpensive index fund, with stocks selected by algorithms, had already grown in size, with assets under management eclipsing traditional active funds in 2019. human participation. The flagship fund of Renaissance Technologies, the first quant outfit set up in 1982, generated average annual returns of 66% over the decades. Faster cables in the 2000s gave rise to high-frequency market makers, including Citadel Securities and Vertu, capable of trading stocks by nanoseconds. New quant outfits like AQR and Two Sigma beat the returns of humans and gobbled up assets.

By the end of 2019, automated algorithms took both sides of the trades; More frequently high-frequency traders faced off against quant investors who had automated their investing processes; The algorithm manages the majority of investors’ assets in passive index funds; And all of the largest, most successful hedge funds used quantitative methods, at least to some degree. The traditional types were throwing in the towel. Philippe Jabre, a star investor, blamed computerized models that had “inexplicably replaced” traditional actors when he closed his fund in 2018. As a result of all this automation, the stockmarket was more efficient than ever. Nothing individuals can invest savings for a fraction of the money on the dollar.

Machine learning promised even more fruit. The way one investor described it was that quantitative investing began with a hypothesis: that momentum, or the idea that stocks that have risen faster than the rest of the index, will continue to do so. This hypothesis allows individual stocks to be tested against historical data to assess whether their value will continue to rise. In contrast, with machine learning, investors can “start with data and look for hypotheses”. In other words, algorithms can decide what to choose and why to choose.

Yet the great march of automation hasn’t been continuous—humans have fought back. By the end of 2019, all major retail brokers, including Charles Schwab, E*TRADE and TD Ameritrade, had reduced commissions to zero due to competition from a new entrant, Robinhood. After a few months, due to pandemic boredom and stimulus checks, retail business began to boom. This peaked in the frenetic early months of 2021, when day traders, coordinating on social media, piled into unpublished stocks, driving up their prices. At the same time, many quantitative strategies seemed to be stalling. Most of the volume underperformed the markets as well as human hedge funds in 2020 and early 2021. AQR closed a handful of funds after continuous outflows.

When the markets turned in 2022, many of these trends were flipped. The retail side of the business fell back as losses piled up. The Quants are back with a vengeance. AQR’s longest-duration fund delivered a return of 44% even though the market dropped 20%.

This zigzag, and the growing role of robots, holds lessons for other industries. The first is that humans can react to new technology in unpredictable ways. The falling cost of trade execution seemed to power the investment machines – until the cost dropped to zero, at which point it fueled a retail renaissance. Even if the share of retail in trading is not at its peak, it remains high compared to pre-2019. Retail trades now make up a third of the trading volume in stocks (excluding market makers). His dominance of stock options, a type of derivative bet on stocks even better,

The second is that not all technologies make markets more efficient. One of the explanations for AQR’s period of poor performance, argues the firm’s co-founder, Cliff Asness, is how extreme valuations became and how long there was a “bubble in everything”. Partly this could be a result of over-zealousness among retail investors. Getting information and getting it quickly doesn’t mean processing it well,” Mr. Asnes says. “I think things like social media make marketers less, not more, efficient … people per- Don’t listen to opinions, they listen to their own opinions, and that can lead to some dangerous craziness in politics and some really weird price action in the markets.”

The third is that it takes time for the robot to find its place. Machine-learning funds have been around for a while and tend to outperform human competitors, at least a little. But they haven’t amassed huge wealth, as they are a tough sell. After all, few people understand the risks involved. Those who have dedicated their careers to machine learning are acutely aware of this. As Greg Bond of Man Numeric, the quantitative arm of Man Group, reports, “To build trust, we’ve invested heavily in convincing clients why we think machine-learning strategies are doing what they’re doing.” are doing.”

There was a time when everyone thought the quants had it figured out. This concept is not there today. When it comes to the stockmarket, at least, automation hasn’t been the winner-take-all event that many fear elsewhere. It is like a tug of war between humans and machines. And although the machines are winning, humans haven’t given up yet.

Sign up for more expert analysis of the biggest stories in economics, finance and the markets money TalksOur weekly subscriber-only newsletter.

© 2023, The Economist Newspaper Limited. All rights reserved. From The Economist, published under license. Original content can be found at www.economist.com