Fat Tail Risk: Why Stock Market Crashes Happen More Often Than 'Normal' Predicts
4 min read
Loading...
4 min read
On 3 June 2026, the IHSG fell 4.11% in a single day, closing at 5,941.07, its lowest level in about five years. Ask a standard risk model beforehand how likely a day like that was, and the answer comes back close to "shouldn't happen." That gap between what the math predicts and what markets actually do has a name: fat tail risk.
Most everyday risk tools, including standard deviation, Sharpe ratio and Value at Risk, quietly assume daily returns follow a bell-shaped curve called a normal distribution. Standard deviation itself just measures how spread out daily returns typically are around their average. Under a normal distribution, about 68% of days should land within one standard deviation of that average, about 95% within two, and 99.7% within three. A three-standard-deviation day is supposed to be rare enough that it shows up only a handful of times in a thousand trading days, and each time should feel like a genuine surprise.
Fat tail risk is what happens when reality doesn't cooperate with that curve. Instead of extreme moves fading out smoothly at the edges, actual market returns keep a heavier tail: big up and down days keep showing up more often than the bell curve says they should. One way to measure this is kurtosis, a statistic describing how heavy a distribution's tails are compared to a normal curve, which has a kurtosis of 3 by definition. Research building on mathematician Benoît Mandelbrot's 1963 work and economist Eugene Fama's 1965 follow-up has found daily S&P 500 returns carry excess kurtosis of roughly 10 to 30, far heavier-tailed than a bell curve. A separate look at daily moves on France's CAC 40 index found the same gap in plain probability terms: a 4% to 5% single-day move actually happened about 0.38% of the time in real trading history, roughly three times more often than the 0.12% a normal distribution would predict.
The starkest historical example is 19 October 1987, Black Monday. The Dow Jones Industrial Average fell 22.6% in one trading session, still the largest single-day percentage drop in its history. Measured against the standard deviations markets were using at the time, published analyses of that day put its odds anywhere from roughly one in ten million to effectively impossible under a normal curve, depending on the exact method used. Whichever estimate you take, a loss the model treated as close to unthinkable happened on one specific Monday, and history kept adding to the list afterward, including October 2008 and March 2020.
You don't need a crash the size of 1987 to see fat tail risk close to home. On 9 March 2020, at the start of the Covid-19 selloff, the IHSG fell 6.58% in a day and triggered multiple trading halts. More recently, the 4.11% single-day fall on 3 June 2026 dragged the index to its lowest close since 28 June 2021, with all 11 IDX sectors closing lower that day and the basic materials sector alone down 9.05%. A day like that is exactly what a model built around two or three standard deviations is designed to call unlikely, and it keeps happening anyway.
Fat tail risk is not a flaw in a single number. It sits underneath the whole toolbox, because Sharpe ratio is standard deviation with extra steps, and standard deviation is the same normal-curve assumption discussed above. A portfolio can carry a reassuring Sharpe ratio right up until a day the underlying return distribution always said was more likely than the model admitted. That is not a reason to throw the ratio out. It is a reason to treat it as one input, not a guarantee, and to look at what a portfolio actually did during its worst real days, not just how spread out its returns look on an average one.
Three practical habits follow directly from fat tail risk. First, stop treating "two standard deviations" as a safety ceiling; it describes a typical bad day, not the worst one that can happen. Second, look at actual historical drawdowns alongside volatility, not volatility alone. The IHSG's real recovery time after its two worst modern drawdowns tells you far more about surviving a fat-tail day than a single standard-deviation figure ever will. Third, size positions and keep enough of a buffer that a day the model calls nearly impossible does not force you to sell at the worst possible moment.
Your own portfolio's volatility and drawdown history are visible on your Portfolio Health view on your NetWort dashboard. Pull it up and check how your worst real day compares to what a plain standard-deviation estimate would have told you to expect.