ookawa/iStockPhoto / Getty Images
Can artificial intelligence be a good investor, and in the process replace fund managers?
Scott Reardon of Dakota Value Funds compiled a list of more than 50 high-profile value investors who outperformed their benchmarks and/or the markets before and after fees over the time during which they were active money managers. They ranged from John Maynard Keynes (over 24 years) and Prem Watsa (over 30 years) to Walter Schloss (over 49 years) and Peter Cundill (over 35 years). And then there was Seth Klarman (over 25 years) and Howard Marks (over 22 years) and both Charlie Munger and Warren Buffett (over 60 years). The key characteristics of all those investors were patience, discipline and long-term perspective.
It is not high IQ and sky high intelligence. It is key understanding of human nature and of institutional biases, as Mr. Buffett likes to say.
While AI most definitely has high IQ, can it replace the human frame of mind needed to make the right investment decisions? Will it be able to avoid making impulsive decisions based on the most recent news, or behave in a contrarian way, which is another key characteristic of value investors?
Those who think that financial markets are natural phenomena will answer these questions in the affirmative. But financial markets are not natural phenomena, they are made by people whose task is managing risk and whose efforts to do so affect risk itself. Financial markets are not like a game of roulette, where what we observe around us does not affect the odds. They are more like a game of poker, where what we observe around us affects the odds.
AI fears sink Stantec and WSP shares, turning the high-flying engineering firms into TSX laggards
And it is hard not to quote Charlie Munger here when it comes to efficient market theory – the belief that stock prices reflect all publicly available information and that price and a security’s value are always the same.
This is what he said following a Q&A after the 2017 Daily Journal annual meeting: “Warren and I have had some effect on investing and thinking, but they are still teaching the Efficient Market Theory at business schools. The old ideas die hard. [Business professors] think that market efficiency is inevitable like physics. Now what kind of nut would want to make stock markets like physics? It isn’t like physics.”
Nassim Taleb, the author of the popular book Black Swan, wrote in the Harvard Business Review in 2009 about the six mistakes risk managers make. One was studying the past; another one was putting emphasis on standard deviation.
If markets are not natural phenomena or a game of roulette, it is difficult to see how AI will replace fund managers. Studying the past and using formulas to replace fund managers will not work. The future is unpredictable and using the past to forecast the future is not the same as having the ability to read between the lines and anticipate what effect events we did not know we did not know will have on the markets. As Howard Marks says, great investors are strongest where AI is weakest, “in dealing with novel developments where there is not enough prior experience for dependable patterns to have been complied (and learned by AI during its training).”
Tech stocks are partying like it’s 1999 – and investors should be wary
Recent evidence also points out the weaknesses AI has in dealing with non-deterministic financial markets. For example, a recent working paper published by The National Bureau of Economic Research tracked the performance of AI-managed portfolios over the past year. The authors tested models based on ChatGPT, Claude, Gemini and Grok. They found that most of the systems lost money. AI models traded too much, recommended stocks based on how much media attention firms received and made very different decisions when given identical instructions.
Moreover, AI has no skin in the game – it is taking risky positions under the idea that the higher the risk the higher the return, even though academic studies show that historically, lower risk led to higher returns. And so, evidence shows that AI seems to choose to invest in higher beta stocks. In fact, AI behaves like a fund manager whose performance has only upside, with limited liability if things go wrong. But successful investors avoid such typical agency problems by investing most of their wealth along with that of their clients. As Howard Marks, who made a fortune in buying distressed securities, said, “The best investors sense potential risk intuitively, and this contributes greatly to their success.”
I tell my students not to fear AI, but they must have three things.
First, reasonable intelligence – which, to be studying at Ivey, they must have. Second, sound principles of operations – which they learn taking value investing at Ivey. And third, the right temperament and frame of mind in making decisions – which they must work on by developing strategies to deal with the weaknesses of human nature and institutional biases.
George Athanassakos is a professor of finance and holds the Ben Graham Chair in Value Investing at the Ivey Business School, Western University. He is the author of the recent book Value Investing: From Theory to Practice.