Sample Size Neglect

Stock investments are supposed to be made based on rational choices. In this module, we have so far learned that investors are not exactly rational and tend to become emotional a lot of times. However, it is important to note that there are some cases in which investors are pretending to be rational but are actually acting on impulses. This means that they are pretending to be following a rational process. However, they are following it haphazardly, and hence the result is not reliable. The sample size neglect is a manifestation of this pseudo-rational methodology in play. In this article, we will understand what sample size neglect is and how it impacts the decision making of investors.

What is Sample Size Neglect?

Scientific studies use samples in order to draw conclusions. This is done because it is not possible to study the entire population. Hence, the sample size needs to be drawn. However, it is the job of the researcher to ensure that the sample actually represents the population. If that is not the case, then the conclusions drawn from the sample are basically null and void.

This is exactly how sample size neglect works in real life. Investors who neglect sample sizes believe that they are following the scientific process. However, in reality, they are just following impulse. Let’s understand this better with the help of real-life examples.

Real-Life Sample Size Neglect Examples

Person A has been making investments with a 50% success rate. This means that one out of the two stocks which they pick is a success. However, Person A meets Person B, who claims that he has an 80% success rate. Person A finds it hard to believe and hence asks Person B to pick three stocks for them over the next month. To their surprise, Person B picks all three successful stocks. This convinces Person A to abandon their investment technique and start following Person B.

Now, it needs to be noticed here that Person A did not directly believe in Person B’s philosophy. They did test it to some extent. However, the test was flawed because the sample size was too small. Three investments in a month could have been a stroke of luck! It could have been because Person B is tracking certain companies or industries which are in vogue at the moment, and hence their stocks have been rising in the recent past. If sometime in the future, this industry or company stops rising, then the predictions will start failing as well.

The reality is that the process followed by Person A is so flawed that it has become meaningless. In order for the experiment to hold any validity, the sample size must have been large. Instead of three stocks, Person B should have picked thirty stocks. Similarly, instead of one month, the period taken into consideration should have been a year or more. Also, ideally, the experiment should be conducted in different types of macro markets, such as the bull market and the bear market. Only if person B’s track record is consistent over different periods of time and in different external circumstances can Person A state with confidence that Person B’s track record is indeed better and hence must change his investment philosophy.

However, this is not what happens in real life. We all know many people who start blindly trusting an advisor and making investments on their advice after their suggestions have turned right a couple of times.

Time Diversification: Time diversification is an important concept related to sample size neglect as well. If an investor believes that a particular investment is good, they must not unnecessarily get distracted by the incoming price data. Taking a small sample size of a data set spread over a couple of months and selling an underlying good investment would obviously be a bad decision. This is where the concept of time diversification should be followed. Time diversification states that if investments are made in risky assets for a longer period of time, some of the risks are automatically canceled out during the long time frame. Hence, investors should not make decisions based on a small sample size.

Gambler’s Fallacy: Gambler’s fallacy is another related concept. Gambler’s fallacy refers to the false belief that outcomes of probabilistic events will behave predictably in the long run. For instance, if we were to do a million coin tosses, the result would most likely be 50% heads and 50% tails. However, the 50-50 ratio does not hold over small data samples. For instance, if only ten coin tosses are done, it is likely that all ten of them will turn out to be heads! The gambler’s fallacy would be to expect five tails after five heads in the short run. The probabilities hold in the long run but not in the short run. Many investors use the same philosophy to bet on stocks that have been hammered in the recent past. This is because they wrongly believe that these stocks have high growth potential.

Following the scientific method in the case of finance is difficult since it requires data from over several years to be used in the process. However, that is indeed the right process that needs to be followed. The inability to do so would create sample size neglect and an illusion of due diligence when, in fact, decisions are being made impulsively.