Artificial intelligence (AI), using evolutionary computing and deep learning techniques, is beginning to be used for trading stocks, commodities, currencies and bonds. The idea is to keep trading rational by having a computer program decide on which trades to engage in so that human emotion is kept out of the process.
John Maynard Keynes, in the aftermath of the Great Depression, had coined the term “animal spirits” to describe human emotion in the financial markets. Since the Great Depression of 1929-39 in the United States – which began with the stock market crash on Black Tuesday, October 29, 1929 after stock prices started falling around September 4, 1929 – many financial crises struck the U.S and world economies: the Black Monday, October 19, 1987 crash, the bursting of the technology bubble in 2000, and the housing bubble in 2007-2008 are all examples of the action of animal spirits in finance. Human emotion not only leads to financial panics and crashes out of fear but also causes bubbles to form because of euphoria. While Keynesianism, or expansionary fiscal policy, became the standard doctrine of economic policy to recover from the Great Depression, nearly a century later, a radical monetary policy experiment of negative interest rates is underway to deal with the Great Recession of 2007-2008 and the subsequent economic recovery that began in 2009.
Collective euphoria and fear are what behavioral economists call ‘herd behavior’ among financial market participants. Herd behavior is the tendency of individuals to mimic the rational or irrational actions of others, though independently they may not necessarily make the same choice. Social pressure of conformity and the belief that a large group cannot be wrong lead to herd behavior. The tech bubble in the late 1990s was formed because venture capitalists invested in startups with unsound business models because everybody was doing the same. That bubble burst in 2000. The housing bubble from 2002-2007 happened because of unsound lending practices to boost the housing market, again because everybody was doing the same. The housing crisis was worse because the government also tacitly encouraged such behavior with policies to promote housing ownership among the population.
It is clear that automated trading systems (ATS) could neither contain the euphoria nor stanch the fear during the 2007-2008 stock market collapses. Then, the question arises if the new generation of AI-ATS can prevent euphoria and fear from taking hold. Can AI-ATS learn in real time that the market mood is euphoric or is besieged by fear by discerning a pattern in current data when compared to past data and neither act in euphoria nor out of fear? Or will AI-ATS choose to ride the euphoria, sense a turning point in the market at the top and then exit to maximize the trading profits for its human owners? Will there be euphoria and fear at all if a majority of trades in the market are performed by AI-ATS and human beings are out of the picture? Moreover, if AI-ATS are available not just to institutional investors but to a majority of retail, day-to-day investors, will market returns, by the law of averages, come down because the trading opportunities are arbitraged out quicker than when human traders are engaged?
AI-ATS are still new and their inventors are quite thoughtfully keeping them discreet to preserve the competitive advantage of their clients. The proliferation of AI-ATS to replace human traders has some ways to go. If AI-ATS agents become common place, it is indeed likely that the financial markets could be less volatile and more stable. But that is yet to be seen.