On Investing Biases and Staying Ahead of the Machine - DecisionBoundaries

On Investing Biases and Staying Ahead of the Machine

I’m an engineer by background and most everyone who works with me knows me as a rationalist: someone who embraces the theory that when we are presented with a variety of options, we will choose the option that both maximizes our individual satisfaction and falls in line with our long-term goals.

But, as 30 years either trading, investing or overseeing trading desks taught me, being a rationalist does not automatically make one rational. In fact, I have bought on upward momentum, added to a sliding position, doubled down when the slide became a crash, and cried for the IMF, the Fed, or another deity to come to the rescue more times than I care to remember.

Now, as an advisor, I enjoy the luxury of perspective and rely on behavioral economics to identify the biases that cause traders and investors to perform below their full potential. Ironically, behavioral economics is such anathema to rationalists like me that Nobel laureate Daniel Kahneman was famously motivated to state: “it seems that traditional economics and behavioral economics are describing two different species.”

My observation is that traders’ and investors’1 biases can be generally clustered around five broad areas:

1 While investors in hard-to-mark securities, like private equity, can arguably mark-to-model their way out of this one bias, they are susceptible to the trappings of the remaining four, since the value of private equity securities can be (imperfectly) hedged, the securities can be sold on one of the now-proliferating secondary platforms, or losses assumed through the inherent dilution from the failure to participate in a misguided investment’s capital increases.

1) The Need for Anchors

When confronted with decisions, it is human nature to begin with the familiar and use it to make judgments. Kahneman and Tversky ran an experiment where they used a wheel of fortune with numbers from 1 to 100 to illustrate this point. With a group of subjects, they spun the wheel to get a number and then asked the subjects numerical questions about obscure percentages like the percentage of ancient Egyptians who ate meat, for instance. The subjects would have to guess whether the right answer was higher or lower than the number on the wheel and then provide an estimate of the actual number. Kahneman and Tversky found that the answer given by subjects was consistently influenced by the outcome of the wheel spin. Thus, if the number on the wheel was 10, the answer was more likely to be 15 or 20%, whereas if the number on the wheel was 60, it was more likely to be 45 or 50%. 

Market prices provide a similar anchor with traded assets and an investor asked to estimate the value of a security is likely to be influenced by the market price, with the value increasing as the market price rises (and vice-versa).

2) The Power of the Story

For better or worse, human actions tend to be based not on quantitative factors but on storytelling. People tend to look for simple reasons for their decisions and will often base their decision on whether these reasons exist. In a study of this phenomenon, Shafir, Simonson and Tversky gave subjects a choice on which parent they would choose for sole custody of a child. One parent was described as average in every aspect of behavior and standing whereas the other was described more completely with both positive (very close relationship with child, above-average income) and negative characteristics (health problems, travels a lot). Of the subjects studied, 64% picked the second. Another group of subjects was given the same choice but asked which one they would deny custody to. That group also picked the second parent. 

While the results seem inconsistent, they suggest that investors are more comfortable with investment decisions that can be justified with a strong story than one without.

3) Overconfidence and Intuitive Thinking

We humans tend to be opinionated about things we are not well informed about and to make decisions based upon these opinions. Fischhhof, Slovic and Lichtenstein generalized the finding by asking people factual questions and found that people gave an answer and consistently overestimated the probability that they are right. In fact, they were right only about 80% of the time that they thought they were. What are the sources of this overconfidence? One might just be evolutionary. That confidence, often in the face of poor odds, may have been what allowed us to survive and dominate as a species. The other may be more psychological. Human beings seem to have a propensity to hindsight bias, i.e., we observe what happens and act as it we knew it was coming all along. 

Thus, there are investors who claim to have seen the 2008 crash coming during earlier years, though nothing in their behavior suggests that they did (the only one who cogently documented its advent was my friend Nouriel Roubini, not an investor).

4) Herd Behavior 

Our tendency to be swayed by crowds has been long documented. In a fascinating experiment, Asch illustrated this by putting a subject into a group of people, asking them a question to which the answer was obvious, and then inducing other people in the group to provide the wrong answer deliberately. Asch noted that the subject changed his answer one-third of the time to reflect the incorrect answer given in the group. While Asch attributed this to peer pressure, subsequent studies found the same phenomenon even when the subject could not see or interact with others in the group. This would suggest that the desire to be part of the crowd is due to more than peer pressure.

While there is a tendency to describe herd behavior as irrational, it is worth noting that we can have the same phenomena occur in perfectly rational markets through a process called information cascade. 

Consider the example of two restaurants, where people come into town one after another. Assume that the first person to come in picks the first restaurant and assume that the choice is random. The second person who comes into town will observe the first person sitting in the first restaurant and is more likely to pick the same restaurant. As the number of subjects entering the market increases, you are likely to see the crowd at the first restaurant pick up, while business at the second restaurant will be minimal. Thus, a random choice by the first customer in the market creates enough momentum to make it the dominant restaurant. 

The same behavior occurs in investing. All too often, we pile up into specific assets and expose ourselves to the vulnerabilities of a crowded trade. 

5) Unwillingness to admit mistakes

It may be human to err, but it is just as human to claim not to err. In other words, we are much more willing to claim our successes than we are willing to face up to our failures. Kahneman and Tversky noticed that subjects, when presented with alternatives to the status quo, often made choices based upon unrealistic expectations. They noted that a person who has not made peace with their losses is likely to accept gambles that would otherwise be unacceptable to them. 

In investing, this is called the disposition effect, i.e., the tendency to hold on to losers too long and to sell winners too soon. In my experience, this is the most widespread and pernicious of investing biases. A review of my own desk’s trading records supports the hypothesis that, for a given period, investors take a statistically-significantly lower percentage of their losses than of their gains. The same sample also reveals that investors tend to sell winners too soon, with winning securities that get sold continuing to outperform for months after the sale. 

I empathize with the high of winning and the pain of losing. I experienced them myself more often than I care to admit. But, in order to succeed, investors need to numb themselves to those stimuli.

Fortunately for the current cohort of investors and traders, devolving their decision-making to machines is not (yet) the optimal response to human biases. The lackluster performance of quant funds is symptomatic of the fact that even our most sophisticated machine learning algorithms are also plagued by bias (chief among them, but not only, overfitting) as are data sets (especially “alternative” ones).

My heuristic today is the same one I taught my traders years ago: lean forward! This was a skiing analogy intended to convey that a counterintuitive behavior when going fast down a mountain will result in fewer falls, just like taking losses often may bruise our ego in the short run but will ultimately result in better performance.



Subscribe to Blog

Leave a Reply

Your email address will not be published. Required fields are marked *