Volatility, and why the window matters more than the number

Realised volatility is calculated from prices that have already occurred. It is a historical measurement, and it is the kind this site publishes. Implied volatility is derived from the prices of options and represents what the market expects future volatility to be. The two are different quantities: one is a fact about the past and the other is a forecast embedded in a traded price. A reader who sees a volatility figure quoted should establish which of the two is meant before drawing any conclusion from it.

The distinction matters because the two can diverge sharply. Implied volatility rises when the market is frightened, often before any large move has occurred, and falls when it is calm. Realised volatility rises only after the moves have happened. A market can therefore show high implied volatility and low realised volatility for weeks, which is a statement about the cost of insurance rather than about what the price has done.

There is also a measurement choice within realised volatility that is rarely stated. The figure depends on the frequency of the returns used — daily, weekly, or monthly — and on whether the calculation uses closing prices or intraday extremes. A volatility figure computed from daily closes will differ from one computed from intraday ranges over the same period, sometimes substantially, because the intraday measure captures moves that the closing series smooths away. A volatility number without its method is not a measurement.

Bitcoin's realised volatility has been high throughout its history and has declined over time. The earliest years, when the market was thin and venues were few, produced annualised volatility figures that would be extraordinary for any conventional asset. The most recent years have produced figures that are still high by the standards of equities or currencies and considerably lower than the asset's own early record. The direction of the change is clear; the magnitude depends heavily on the window chosen.

The decline has two causes, and they are worth separating. The first is that the market has grown. A larger market absorbs the same flow with a smaller percentage move, so the same news produces a smaller price change than it once did. The second is that the participant base has changed. A market dominated by retail speculation behaves differently from one with institutional participants, derivatives for hedging, and regulated venues. Both changes reduce measured volatility, and both are structural rather than cyclical.

The practical consequence is that a volatility figure from the early record is not a good guide to the asset's current behaviour, and a figure from a calm period is not a good guide to its behaviour during stress. Volatility clusters: quiet periods are followed by quiet periods and violent ones by violent ones, so a low reading today says nothing about tomorrow. A reader should treat any single volatility figure as a description of a specific window and should look at the drawdown record for the realised downside that volatility is meant to summarise. The risk and volatility page publishes the site's own volatility series with its method stated.