Skewness and Asymmetry
Skewness describes asymmetry in a distribution. It matters when a pattern of frequent modest outcomes includes rare, much larger outcomes on one side; a typical result or an average alone may conceal what matters about the tail.
THE IDEA
SKEWNESS AND ASYMMETRY
Skewness measures asymmetry in a probability distribution. For a distribution such as the normal curve, the two sides are symmetric and skewness is zero. Positive skew is associated with a longer or heavier right tail, and negative skew with a longer or heavier left tail. Those directions do not, by themselves, specify whether most observations are gains or losses.
In a return distribution, frequent small gains can coexist with rare large losses, or frequent small losses with rare large gains. Describing a position simply as bullish or bearish says little about that balance of probabilities and payoffs.
A few extreme observations can strongly affect an average. If the historical sample has not captured the relevant extremes, it can also give a misleading picture of the distribution. This is the connection to epistemic opacity: uncertainty about what the data allows us to know.
When to use it
Evaluating investment portfolios by analyzing return distributions for negative skewness (tail risk).
Designing insurance policies or safety margins for rare events like natural disasters.
Understanding customer behavior when most customers buy little but a few buy a lot (positive skewness in spending).
Comparing probabilities and possible gains or losses before relying on a simple bullish or bearish label.
FROM THEORY TO EVERYDAY LIFE
In practice
Financial returns with negative skewness
Consider a strategy that produces small monthly gains but occasionally suffers a much larger loss, as might occur during a crisis such as 2008. A record dominated by winning months can distract from the size of the downside; both probabilities and payoffs matter.
Weather extremes and positive skewness
Imagine daily rainfall in a desert: many days are dry, with occasional heavy downpours. Mean rainfall can be useful, but it does not show the size or frequency of the downpours needed for flood planning.
A MOMENT TO REFLECT
Put it to work.
In what areas of your life or work do you implicitly assume symmetry when the underlying distribution is actually skewed?
How can you protect yourself from the 'vicious child'—a rare but devastating outlier—in your decision-making?
Why might historical data (e.g., stock returns) be insufficient to estimate skewness accurately? What does this say about epistemic opacity?
CONTINUE THE CONVERSATION
Source texts
- Fooled by Randomness The Hidden Role of Chance in the Markets and in Life (Nassim Nicholas Taleb)