In order to derive meaning from life experiences, people have developed an innate propensity for classifying objects and thoughts. When they confront a new phenomenon that is inconsistent with any of their preconstructed classifications, they subject it to those classifications anyway, relying on a rough best-fit approximation.
There are two main types of representativeness bias, namely (i) base-rate neglect and (ii) sample-size neglect. We will focus on the latter, since it occurs more frequently in trading.
In sample-size neglect, traders, when judging the likelihood of a particular trade outcome, often fail to accurately consider the sample size of the data from which they base their judgments. They incorrectly assume that small sample sizes are representative of populations. This is also known as “the law of small numbers”.
This problem is observed when traders try to backtest systems by using small sample sizes of data, and extrapolate their favourable results. However, these results are most likely not representative of the effectiveness of the system. This is a common tactic applied in marketing gimmicks.
Another common phenomenon has to do with hot tips. For example, you might hear someone say “my broker gave me three great stock picks over the past month, and each stock is up by over 10%”. While this is enough to sway most people, thinking that the broker is a genius, this assessment is based on a very small sample size.
What is the best solution for this?
If you want to evaluate the effectiveness of system or the stock-picking skills of a person, make sure you do it over a large sample size, and count both the hits and misses. This will give you a more complete representation of reality.