If you look at a long price history and ask when the large daily moves happened, they will not be scattered evenly. They arrive in bursts. Quiet stretches follow quiet stretches, and violent days follow violent days. This is not specific to crypto — it shows up in most traded markets — but it is unusually pronounced here, and it has practical consequences.
Why bunching happens
Three mechanisms do most of the work, and none of them requires anything mysterious.
The first is that information arrives unevenly. Long periods pass in which nothing much is learned, then something is learned and a great deal of repricing happens at once. Markets do not receive news on a smooth schedule, so they do not reprice on one either.
The second is leverage. When positions are financed, a move against them can force them to be closed, and closing them pushes the price further in the same direction, which forces more closures. That feedback loop is mechanical rather than emotional, and it is why the largest single-day moves in crypto so often occur in the direction of the crowded position rather than the direction of the news.
The third is liquidity. The depth available to absorb an order is not constant. It thins out precisely when volatility rises, because the people providing it widen their quotes to protect themselves. The same order that would barely register on a calm day moves the price meaningfully on a turbulent one. Our glossary entry on liquidity covers the mechanics in more detail.
What this means for reading a chart
A quiet week is not evidence that the asset has become less risky. It is evidence that the asset has been quiet, which is a statement about the recent past. Volatility is persistent in the short run — calm tends to be followed by calm — but persistence is not permanence, and the transition from one regime to the other is usually abrupt rather than gradual.
This is also why a single day’s percentage move is a poor summary of anything. A five per cent day means something quite different in a stretch where the typical day is one per cent than in a stretch where the typical day is six. Context is not decoration here; it is the whole measurement.
Sentiment is a symptom, not a cause
Sentiment indices are often read as though they lead the market. It is more defensible to treat them as a description of where the market has recently been. Fear readings tend to be high after prices have fallen, which is not a prediction so much as an observation with a lag. We publish one such measure on the Fear and Greed page, and we would rather it be read as context than as a signal.
The honest limit
Knowing that volatility clusters does not tell you when a cluster starts. That is the part the model does not give you, and anyone selling the transition as predictable is selling something. What clustering does give you is a better prior: it makes a sudden violent day less surprising, and it makes a calm month less reassuring than it feels.
For the wider cycle framing, see our guide to bull and bear market cycles, and for how to read the charts themselves, how to read a crypto chart. Nothing here is financial advice, and none of it is a forecast.