3. Fat tails and skewness

Level 2: Distributions and why markets are not "normal"

On October 19, 1987 the S&P 500 fell more than 20% in a single session.

With the bell curve from the previous module in hand, that day should not have existed. It is not that it is 'rare': the model says it should never have happened, in the entire history of organized markets.

It happened. And the real histogram of any index carries the mark of why.

Three systematic differences

Build the actual histogram of daily returns of any broad equity index, such as the S&P 500, over several decades. Compare it with a normal curve having exactly the same mean and standard deviation.

You will notice three systematic differences:

• There are more days very close to zero than the normal predicts: a sharper peak in the center.
• There are fewer days with moderate moves, neither small nor extreme.
• There are more days with extreme moves — both upward and, above all, downward — than the normal considers plausible.

This last feature is what people colloquially call fat tails: the extremes of the real distribution carry more weight than those of a comparable normal.

Kurtosis: tall peak and fat tails

The technical name for this phenomenon is kurtosis.

You can grasp it without any formula. It is a measure of how much probability piles up in the center and in the extremes of a distribution, compared with the normal, at the expense of the intermediate zone.

A distribution with kurtosis above the normal is described as leptokurtic. It has exactly that tall peak and fat tails pattern: it looks more sharply pointed at the center than the normal bell, but its extremes decay toward zero far more slowly.

Daily financial returns are, almost universally, leptokurtic. There are more quiet days than expected, more catastrophic days than expected, and fewer days of intermediate volatility.

Skewness: the left tail is longer

The second systematic difference is skewness. It measures whether a distribution is symmetric, like the normal, or has a longer tail on one side.

The returns of stocks and equity indices show, fairly consistently, negative skew: the left tail, that of losses, is longer and heavier than the right tail, that of gains.

In practice this means that extreme downward moves tend to be sharper and more sudden than extreme upward moves. Compare the two sides:

• A market can rise 20% gradually over several months. But it rarely rises 20% in a single session.
• By contrast, it has fallen more than 20% in a single day, on October 19, 1987. And it has fallen more than 9% in a single session on several historical occasions.

Why the market is asymmetric

There are underlying economic reasons for this asymmetry. It is not a mere statistical accident.

Fear reacts faster and more forcefully than greed. Forced selling driven by margin calls, the liquidation of leveraged positions, and the panic of sellers who need to get out now, not tomorrow, generate abrupt falls.

Those falls have no symmetric equivalent on the upside, simply because nobody is ever forced to buy with the same urgency.

There is a second driver. Many strategies and instruments have an asymmetric payoff profile by construction, such as selling options or certain leveraged strategies. They earn a little frequently and, occasionally, lose a lot, which amplifies that heavy left tail in aggregate market data.

The Black Monday arithmetic

Return to the example from the opening, because the figures deserve a slow look.

On October 19, 1987, so-called Black Monday, the S&P 500 fell more than 20% in a single session. With a typical daily standard deviation for the index of around 1%, that fall represents a move of roughly 20 standard deviations.

Under a strictly normal distribution, the probability of such an event is so close to zero that, in practical terms, the model says it should never have happened. Not in all the time since organized markets came into existence.

And it was not the only time moves the normal would call impossible have been observed. The years 1997, 2008, 2010, 2015, 2020 and 2025 supply further examples. Each with its own trigger, and all with the same statistical signature: far more extreme than the Gaussian bell allows.

The statistical fingerprint of black swans

This gap between normal theory and empirical reality is not a minor technical detail. It is the statistical fingerprint left by what the market theory course called black swans.

You are not going to dig into the causes and practical implications of those events yet: that comes in the final module of this level.

But the connection is worth making clear right away. Fat tails and negative skew show up in any real return histogram. And they are the statistical evidence for an underlying idea: extreme events are not a marginal theoretical curiosity, but a structural and recurring feature of how markets move.

What you miss if you ignore it

For anyone building or evaluating a strategy, ignoring this has very concrete consequences.

A Sharpe ratio computed on a strategy with strong negative skew can look excellent for years, with small, steady gains. And hide a catastrophic loss risk that materializes only on rare occasions, but devastatingly when it does.

It is the pattern colloquially described as 'picking up pennies in front of a steamroller': everything is fine, until it stops being fine all at once.

Recognizing fat tails and skewness in the data is the first step toward not mistaking a quiet run for a genuine absence of risk.