Sunday, June 5, 2011

How We Know What Isn't So: Chapter 2

Value investors believe the market is not perfectly rational. To understand why, an examination of human behaviour is required. In How We Know What Isn't So, which is recommended by a number of value investors and behavioural economists, Thomas Gilovich explores the fallibility of human reasoning. Only by understanding our flaws can we seek to improve on them, thereby ameliorating our decision-making processes.

Humans look for patterns where there aren't any. Several examples of this phenomena are discussed in this chapter, including a detailed examination of the "hot hand" idea pervasive in the sport of basketball.

It is believed that players are streaky, in that if they have made (missed) a few shots in a row, they are more likely to make (miss) their next shot as well. Examining data from the NBA, the authors show that players do not exhibit "hot hands".

And yet there continues to be a belief in the existence of "streaky-ness" in player performance. Gilovich believes there are two likely explanations for this: the clustering illusion and the regression fallacy.

Clustering illusion describes the idea that humans see patterns in small samples of data that are not statistically significant. In small samples, however, there is a decent probability that all or a significant proportion of coin flips will turn up heads, for example. Humans extrapolate such phenomena to mean something more than it really does.

Regression fallacy describes the tendency for humans to ascribe some false explanation where all that is really occurring is normal regression, where regression refers to the fact that when two variables are correlated (e.g. parent's height with child's height), an extreme outcome in one variable (e.g. the parent is very tall) will not be accompanied by such an extreme outcome in the other variable (e.g. the child will not be as tall as the parent). This tendency can make people believe something has caused a sub-par performance following an outstanding performance, whereas in actuality the outstanding performance was out of the ordinary.

Lines of thinking that look for patterns have probably served our species well. We can capitalize on ordered phenomena in a way that we cannot when things are unordered. Over the course of history, the prediction of patterns has led to discovery and advance in all fields of study. As a result of our trying to find the order in things, however, we often force an order when none is justified.

No comments: