As I’ve written before, algorithmic trading has taken over Wall Street…
Today, more than 80% of U.S. stock trades are executed by computers.
Armies of “quants” – quantitative analysts – parse reams of data to discover the formulas behind the most profitable trades.
You can attribute the rise of algorithmic trading to two trends.
First, there has been an explosion of computing power and data.
The average investor now has more information on his smartphone than George Soros had when he “broke the Bank of England” in 1992.
Second, computers have made the world’s stock markets vastly more efficient.
Human stock pickers have always struggled to outperform their benchmark indexes, and beating the market is now more challenging than ever.
Computers can sift through millions of sources with lightning speed. Compare that with a room full of financial analysts sifting through balance sheets and regulatory filings…
Still, Wall Street’s quants remain on the relentless hunt for new strategies to wring more profits from the markets.
Some believe that artificial intelligence (AI) is the key to developing an edge.
So the question arises: Is AI-driven investing a sign of things to come or a fad driven by marketing and hype?
The jury is still out. But history has shown that investing fads rarely pan out. AI investing is unlikely to be any different.
The Problem of Human Irrationality
This past weekend, I got into a heated discussion with a mathematics Ph.D. at a local London pub.
The subject of our debate? It wasn’t the latest football (soccer) game playing on the screens above our heads. It was whether investors are perfectly rational in making their investment decisions.
The man had written a book on options trading. And he argued that – on the whole – investors make rational investment decisions.
I vehemently disagreed.
I pointed out that evolution had not optimized humans to be soulless stock pickers. Instead, our brains are divided into three parts. And only our outer brain – the cerebral cortex – is capable of making rational decisions.
Dozens of cognitive biases also impair human decision-making.
Daniel Kahneman – a Princeton psychologist who had never taken an economics course – won the Nobel Memorial Prize in economics for uncovering the first cognitive biases.
Today, Wikipedia lists more than 100 cognitive biases related to beliefs and decision-making.
Add up the impact of all these biases, and humans – including investors – clearly exhibit all sorts of “irrational” behavior. We buy at the top. We don’t cut our losses. We don’t let our winners run.
(By the time the mathematician and I had finished our last pints, the consensus was that I had won the debate.)
The AI Solution
AI-powered investing offers a solution to the problem of human irrationality.
It replaces human decision-making with a super-rational, super-powerful artificial brain.
Today, at least two fund managers, EquBot and Qraft Technologies, are running AI-powered exchange-traded funds (ETFs).
EquBot has teamed up with IBM’s Watson technology. (That’s the computer that famously beat all-time Jeopardy! champion Ken Jennings at the game.)
Watson scours the news, social media, analyst reports, financial statements and even job postings to drive its investment decisions.
Qraft uses AI to identify historical data relationships with future macroeconomic and financial factors. Then, it invests in those stocks that are likely to produce higher returns quickly.
And Qraft’s AI does seem to make investment decisions you’d never expect from a human.
For example, the Qraft AI-Enhanced U.S. Large Cap Momentum ETF (NYSE: AMOM) had a whopping 14.7% weighting in Tesla (Nasdaq: TSLA) in August 2020.
By September 1, 2020, it had sold out of Tesla completely. It bought the stock again in November, amassing a stake of 7.6% by January of this year. By February 1, it had sold the entire holding yet again.
It’s hard to imagine a human fund manager doing the same.
An Achilles Heel for AI?
So how have these AI ETFs fared?
EquBot’s AI Powered Equity ETF (NYSE: AIEQ) and AI Powered International Equity ETF (NYSE: AIIQ) have beaten their benchmarks since their inception.
Qraft’s AI-powered ETFs have shown similar success since their launches.
That said, these ETFs are far from perfect. Some are lagging behind the plain-vanilla SPDR S&P 500 ETF Trust (NYSE: SPY) this year.
I see several challenges.
First, AI is great at crunching numbers and finding patterns. But it bases its decisions mainly on historical data, and the past does not always equal the future.
In the investing world, rules are not static. They continuously change, even as you are playing the game.
For all AI’s computing power, it’s unclear whether it can keep up. Reality is just too complicated.
Second, AI-powered investing contains the seed of its own destruction. Factors like value and momentum will become far less valuable as AI identifies them.
By making markets more efficient, AI reduces its own likelihood of outperforming over the long term.
The bottom line?
AI has promise. But, no, it is not the be-all and end-all of investing.