The Flawless Investor: Howard Marks on whether AI can beat the stock market

Oaktree Capital founder Howard Marks, known on Wall Street for his research notes on all aspects of investing, from index funds to cryptocurrency, tested AI to write a new text - and was impressed with its abilities. Nevertheless, Marks is confident that chatbots are not yet capable of reaching out to experienced investors in making transaction decisions. What arguments he makes in his February memo - in an abridged retelling by Oninvest.

***

Artificial intelligence can process more data than any investor, remember it better, and recognize past patterns that precede success more effectively. It doesn't have to be fearful or greedy. It is likely to be less likely to be optimistic or pessimistic, to hold on to entrenched beliefs, or to be overly focused on the latest information - unless it is trained to be. AI isn't into fashion fads and isn't afraid to miss a trend that others are chasing.

On the other hand, it also has drawbacks. Great investors are much more than just fast and unbiased data processors. They need to be strong where AI can be weakest - in working with new information, where it lacks the previous experience to form reliable patterns. Great investors have to make subjective decisions, exercise taste and discernment. How can AI do that? Here's something else: AI has no interest in the outcome. It doesn't feel the burden of over-concentrated positions or the fear of losing capital. It doesn't have the risk aversion familiar to humans. And the best investors sense potential risk intuitively, which contributes to their success.

In 2021, I wrote a note called Something of Value about the time my son Andrew and I spent during the pandemic discussing the essence of investing. In the note, I shared an observation Andrew had made that "readily available quantitative information about the current moment" could not be the key to outstanding investing for a simple reason - everyone has that information. To this we should now add the fact that AI is likely to do a better job of processing this information than anyone else. So the prospects for those who want to outperform the market using this information seem very limited.

That's what should be the key to success:

(a) Correctly assessing the importance of the information and its implications,

b) assessment of qualitative factors such as management effectiveness and product innovation and/or

c) predicting the future of companies.

Few have such exceptional insight. Just as index investing has left many active investors unemployed who did not deliver value and did not earn their commissions, artificial intelligence is likely to replace humans who cannot perform tasks (a), (b) and (c) as well as it can.

I'll add one more thing. I see AI as a tool that formulates "hypotheses" about what will work in the future. It can analyze historical data, learn past patterns, and predict future leaders. When investors are confronted with new and untested products, CEOs or industries, one has to rely on "speculation and assumptions" rather than facts. Given that AI is limited in its ability to deal with entirely new situations, can it outperform humans in thinking about new things?

Because much of the investment process comes down to assumptions, and because AI is not entirely reliable, I believe it is unlikely that AI will be a flawless investor. It will offer well-reasoned hypotheses, but like human decisions, they won't always be correct. So before investors make decisions based on AI hypotheses, we need to test whether the hypotheses are valid.

This article was AI-translated and verified by a human editor