AI is now better at picking stocks than humans. Thus spake not Zarathustra but a May 2024 Chicago Booth research paper, ‘Financial Statement Analysis with Large Language Models’. Reaction to this information so far straddles the spectrum from ‘Financial apocalypse is nigh!’ to ‘So what?’ Frankly, it has taken AI a while to pick stocks when it has been, say, filtering drug molecules with consummate ease. And just as in medicine, portfolio management based on richer information than what can be fed into machines remains in human hands. Mature markets make stock-picking less relevant in generating investment returns. So, getting better at it should pull down risk and, as a corollary, returns. But, unlike medicine, markets are essentially a pursuit of the irrational. They will be driven by emotion, no matter the sophistication of the maths.Stock markets have been around since the 17th century. Most of the quantitative tools that go into making investment a ‘science’ have emerged in the interim. Humans have used them to get better at reading balance sheets. Now, machines may be taking over, thank you very much. Yet, markets haven’t been able to tame volatility. It’s in their nature not to remain flat, irrespective of advancement of probabilistic determination. Bulls and bears will be at it no matter the information being priced in.
Could there be a future where only machines trade against machines? That would be a journey down the rabbit hole of diminishing returns. Chasing alpha will remain a human preoccupation, using as much help AI or other techs further down the tube can provide. The gambling instinct is rooted in human evolution, and survival strategies will stop it from crossing the machine barrier if such migration were to become possible. Machines will hone human instinct by making information more freely available, and turning the risk-averse into risk-takers. The field of irrationality may narrow for the markets, but not the intensity. That should leave ample space open for human market intermediaries.
