For decades, Eugene Fama’s efficient market hypothesis (EMH) served as the foundation for all financial market activity. Under this theory, investors make perfectly rational decisions because all relevant and available information is immediately incorporated into stock prices. According to EMH, undervaluation and overvaluation cannot exist because they would represent inefficiencies that would be traded out of the market by informed investors.
The efficient market hypothesis, however, represents utopian market conditions that simply do not live in reality. Real markets have pockets of inefficiency where information and liquidity are relatively sparse. Furthermore, humans are emotional creatures, and investors exhibit psychological responses that cause irrational decision-making. What follows is a list of psychological pitfalls to avoid, and an introduction on how the implementation of artificial intelligence (AI) with respect to investment strategies can do just that.

Underreaction and Overreaction

When responding to new information, investors often miss the mark. One of the reasons EMH does not hold up is that market participants do not respond perfectly to new information. The tendency to doubt and hesitate can result in market underreactions, which result in slow price adjustment. Investors are similarly inclined to overreact to news, especially when it is highly visible in the media. Market overreactions result in prices rising or falling in excess of the true price correction, adjusting later toward the missed target. Both types of reactions result from investor bias and emotion, and both may be tempered by incorporating artificial intelligence into trading decisions, as an AI operates without sentiment. A thinking AI neither underreacts nor overreacts; it only acts on its continuous flow of intel.

Representativeness and Anchoring

Investors often have a difficult time ignoring a past bias in the presence of new information. For example, when a stock has previously performed well or poorly, investors are biased to believe that same performance will continue. While studies show that relative risk aversion does indeed change in the face of a long string of losses, investors hold tight to these beliefs even in the presence of information to the contrary. Research by Richard Thaler and Werner De Bondt shows how clinging to those past beliefs can result in poor investment decisions. In other words, not only could last year’s losers become this year’s winners, but long-term flops can one day turn around as well. As there is no human bias within machine learning, an AI will factor in all data — including investors’ collective approach to behavioral finance and the results that such human biases generate — and will make market predictions accordingly.

Sunk Costs

People are slow to admit their mistakes, and that includes poor investments. When an asset’s price drops, the investor considers whether the loss is short-term in nature or represents a long-term problem with the underlying firm. If a short-term market decline or other issue caused the drop, the investor may choose to hold the asset in anticipation of an eventual market correction. At other times, the loss is due to a fundamental change in the company that reduced its long-term value. Even in these cases, investors are hesitant to sell because doing so involves taking a real loss. Since any loss the investor has is only on paper until the asset is sold, the investor protects their previous choices by holding off too long and experiencing even greater losses. Artificial intelligence does not feel the uncertainty and anguish that occasionally comes with human decision-making. A self-learning AI is constantly consuming vast tracts of data and unearthing non-linear patterns and can provide additional, unbiased forecasts that validate the negative long-term outlook.

Confirmation and Superiority

Rather than seeking out new and diverse opinions, investors sometimes tend to stick to groups of like-minded individuals, or just flat-out think they know better than everyone else. In the former, this can be a problem when seeking investment advice, because the group will simply confirm each other’s opinions; in the latter, investors find themselves believing their information and beliefs about a firm are correct even if analysts say otherwise. Investors seek out this confirmation or self-affirmation when they begin to question their own choices. Rather than solidifying a bad decision, investors really need to embrace a new outlook on the situation. They can use machine learning to provide a fresh, new analysis of market trends to identify appropriate investment strategies.

Conclusion

Whether the argument about rational markets or behavioral finance is correct, both sides can benefit from the addition of AI. Machine learning allows AI to continuously learn from itself and new data so that it can make biased-free predictions. Artificial intelligence doesn’t (yet) have a superiority complex; rather, it simply provides an additional source of input for investors to consider, one free from the biases of mere mortals, and the psychological missteps that often accompany them.

References:

http://www.investopedia.com/articles/investing/060513/avoid-these-common-investing-psychology-traps.asp?optly_redirect=integrated&lgl=vtas-baseline

http://www.institutionalinvestor.com/Article/2485942/Asset-Management/Using-Behavioral-Finance-to-Better-Understand-the-Psychology-of-Investors.html?ArticleId=2485942&single=true#.WcfIba2ZPR0

http://www.morningstar.com/InvGlossary/efficient_market_hypothesis_definition_what_is.aspxhttp://fac.comtech.depaul.edu/wdebondt/Publications/DoesStockM.pdf