In some ways, investing today is no different from investing 300 years ago - strong, cash flow positive businesses do well over long periods of time, bubbles form in the short term, cheap debt generally leads to bad outcomes. These things have all been constants, no matter which period of financial history one looks at.
However, in many other ways, investing today differs significantly from investing in the past. One of the most pronounced changes that has occurred over the past few decades is the gradual computerization and automation of trading and investing. It is, therefore, perhaps unsurprising that many commentators have hypothesized that investing algorithms will remove the need for humans from the market altogether. Might this be true? In a talk with the CFA Society of Chile, value investor Howard Marks (Trades, Portfolio) explained some of the limitations of automated decision making.
Exacerbating trends
Marks has written and spoken extensively on the role of human psychology in the investment process. Like many other value investors, he believes that extreme emotions can cause significant distortions in asset prices, which level-headed individuals can take advantage of. An algorithm has no psychology. Might it therefore offer superior investing results over the error prone human brain, and smooth out the market cycle? Marks doesn’t think so:
“No machines are better than the instructions that they are given. My belief is that many algorithmic approaches are momentum-oriented, and they tend to concentrate capital in the things that have been doing well, in the assumption that they will continue to do well. I think there’s every reason to believe that these approaches that put the decision making on autopilot would tend to perpetuate trends and exacerbate them as long as they last. It’s estimated that over 95% of transactions on the New York Stock Exchange are made by machines nowadays.”
Along with algorithmic investing, there has been a parallel but related development: the rise of indexed investing. Indeed, many algorithms simply follow the benchmark indexes. This comparative crudeness makes it hard to invest based on value. If your criteria for buying a stock is that it is part of an index, then that is a separate consideration to whether or not it is undervalued.
In other words, while a computer may not suffer from the emotional imbalances that a human investor might, it also cannot deviate from its pre-programmed set of rules and, therefore, is likely to contribute to the development of any trends that might be taking place. And as long as excesses continue to occur - and there is no indication that the number of bubbles has decreased over the last few decades - algorithms are likely to exacerbate them significantly.
Read more here:
- The Value Investor’s Handbook: Understanding Return on Capital Employed
- Buffett, Munger Explain the Most Important Metric to Look for When Investing
- Charlie Munger: Know When to Take Advantage of Investment Windows
Not a Premium Member of GuruFocus? Sign up for a free 7-day trial here.
