Volatility Clustering
The day after the market swings hard, have you ever felt like it somehow swings hard again? That's not your imagination. It's a real phenomenon called volatility clustering.
"Large changes follow large changes"
Volatility clustering is the phenomenon where 'periods of high volatility and periods of low volatility each appear bunched together.'
In 1963, mathematician Benoit Mandelbrot put it this way: "Large changes tend to be followed by large changes—of either sign—and small changes tend to be followed by small changes."
In other words, a crash often does not end in a single day. The day after a big drop tends to be followed by a day that either rises sharply or falls sharply—in any case, a day of 'large movement.' Conversely, a quiet market stays quiet for a while.
You can't predict direction, but the magnitude bunches up
There's a very important distinction here. Volatility clustering means that the 'magnitude of the movement,' not the 'direction of the return,' bunches together.
Whether the market will rise or fall tomorrow can hardly be predicted from today's movement (the autocorrelation of returns themselves is weak). But 'whether it will also move a lot tomorrow' can be guessed to some degree (the autocorrelation of the magnitude of movement is strong).
Statistically, returns themselves have almost no correlation with each other, but the absolute value or square of returns shows strong, long-lasting autocorrelation. It's 'random in direction, sticky in magnitude.'
The models that captured this property are ARCH (Engle, 1982) and GARCH (Bollerslev, 1986). Engle received the 2003 Nobel Prize in Economics for this contribution.
What it means for investors
Volatility clustering offers a few practical lessons.
First, the thought 'it was quiet yesterday, so it's safe today' is dangerous. Quiet periods end eventually, and once things start churning, that churn continues for a while.
Second, rushing in during the middle of a crash thinking 'this must be the bottom' can be risky. High-volatility stretches can last not just a few days but weeks or months.
Third, that's why 'an asset allocation that can withstand bunched-up volatility' is more realistic than 'timing that tries to call the direction.' Volatility clustering explains once again why market timing is so difficult.
Preguntas frecuentes
Q. If there's volatility clustering, can I predict the next crash?
You can guess to some degree 'whether volatility will be large or small,' but 'whether it will rise or fall' is still hard to predict. Volatility clustering only tells you 'periods where a large movement is likely'—it doesn't tell you the direction. So trying to use it to call the direction of trades is still risky.
Q. What's the relationship between the VIX and volatility clustering?
The tendency of the VIX (fear index) to stay elevated for a while once it spikes is exactly an example of volatility clustering. When the VIX surges during a crisis and then, instead of dropping back within a few days, stays high for weeks, that too is the same phenomenon.
Páginas relacionadas
📋 Los resultados se basan en datos históricos; las rentabilidades pasadas no garantizan rentabilidades futuras.
📋 Este servicio se ofrece con fines educativos para ayudarte a entender la inversión, no como asesoramiento de inversión.