Computers make around 65% of trades in the stock market. Computers have beaten the best chess players and the best Go players. Why shouldn’t they win at the investment game as well? There is a reason that computer algorithms have not, over the longer-term, beaten the S&P500 with dividends reinvested.
Malcolm Gladwell posited that it takes 10,000 hours of intensive practice to achieve mastery of a complex skills and concepts.1 That is 10,000 hours of human time. Consider the game of chess; in just the first seven moves there are over three trillion potential positions. Assuming humans can play a game of chess in an hour, it would take 10,000 hours to acquire the skills of a master. If a person played one game per day, it would take 54 years to acquire mastery of the complexities of chess. A computer can play thousands of games a day, completing the same task in several days.
That is the advantage of computer learning, or artificial intelligence (AI).
That is typical, in our opinion, of the thinking of those investors who let computer quant programs dictate their stock market strategies. Todd Hawthorne, founder of Frontier Asset Source Technologies, was interviewed by Jeff Praissman for Interactive Brokers. He explained the process and problems of AI this way:
“The decision that the AI makes is based on a desired outcome as opposed to a static quality. So what does that mean? You’re telling the computer what the pro forma outcome that you desire is, as opposed to back testing a bunch of factors that have a correlation to the outcome that you’re looking for. That’s a slightly different but very important distinction in that number one: you can change what that desired outcome is. So, in quant systems we’ve had a lot of times in the market where we call it, trying to put the elephant through the keyhole where every single client fund is constantly running these back tests on all these factors, and one or two factors tend to bubble up to the surface and then everyone is overweight those factors which seem highly correlated to our performance. And then there is a tectonic shift of some sort. Those factors stop working, and then the quant funds have to sell their stocks, and it’s the elephant through the keyhole syndrome and those stocks go down disproportionately because everyone is trying to get out at the same time.”2
Consider a coin toss that produces a sequence of H-H-H-T-T-T-H. You and I know the probability of another head is just 50%, but if computers are reading this without any previous input they can easily misinterpret the data. Because it has only witnessed that single sequence, the machine would assign a high probability, perhaps 80-90%, to the next flip turning up another head. In fact, it would not be that unusual to get that sequence. Flipping a coin will once every 128 times produce that exact sequence. This is just an illustration.
Computer algorithms crunch tons of numbers looking for past patterns called “factors”, and then extrapolating those patterns to find stocks to buy or sell. But markets are not like chess or Go. The rules of chess never change. For centuries, the rules of chess have always been the same. But stock markets are driven by the decisions of millions of people who have millions of different experiences and emotions. The market is dynamic and always changing. Those past patterns change, and, when they do, bunches of computers buy or sell at the same time. It is the elephant trying to go through a keyhole syndrome.
There are patterns which will reoccur, but those are few and far between. Today, the Federal Reserve is raising interest rates to counter inflation; only a few times in the past 50 years has the Fed raised rates. The impact of this on certain types of stocks has driven the algorithms to apply the past to the present. While history may rhyme, it does not exactly repeat. Humans understand that. But the majority of trades are being done by computers, not humans. The example of the H-H-H-T-T-T-H is not far wrong in examining the impact of rising interest rates to cure inflation.
COVID-19 is an example of one of those times when the past betrayed the computer algorithms. “Almost three-quarters of quants surveyed by Refinitiv in October  said that their models had been hurt by Covid-19, and a small but eye-catching minority of 12 per cent declared that their models were obsolete… As Ted Aronson, a value-oriented quant investor recently noted to the Wall Street Journal after shuttering his hedge fund AJO after a dismal performance stretch: “It can all work for years, for decades, until or except when the not-so-invisible hand comes down and slaps you and says, ‘That’s what worked in the past, but it’s not going to work now, nope, not any more’”3
Back to our simple example, when the next flip of the coin turns up tails instead of heads, the computer adjusts the probability to accommodate H-H-H-T-T-T-H-T. The elephants head for the keyhole.
Bear markets have been around for as long as we have had data. What is new today is the reaction of computers grafting past data onto the present. They sell as fast as possible, a herd of elephants stampeding through that keyhole. As companies report their earnings, the computers will begin to adjust and reverse their actions.
We believe what is important to understand now is the opportunity to be in those stocks that will be the target of computer buys in the future.
1. Gladwell, Malcolm. 2009. Outliers. New York, NY: Back Bay Books.
2. Praissman, Jeff, and Todd Hawthorn. Rise of the Six-Million Dollar Analyst, How Artificial Intelligence (A.I.) is Changing the Investment Landscape. IBKR Podcasts Episode 39, September 20, 2022, https://www.tradersinsight.news/ibkr-podcasts/rise-of-the-six-million-dollar-analyst-how-artificial-intelligence-a-i-is-changing-the-investment-landscape.
3. Robin Wigglesworth, “A Terrible, Horrible, No-Good Year for Quants” (Financial Times, November 3, 2020), https://www.ft.com/content/d59ffc34-5a34-4cdd-bbbf-5a0e82859f1c.