Volatility index

An index calculated from option prices rather than from the price of anything traded. The best known examples are published by options exchanges on large equity indices, and the method is broadly shared: prices are taken across a strip of strikes at two nearby expiries, combined into a single measure of the variance the market is paying for, and interpolated to a constant horizon so that the published number describes the same window every day. Because the calculation uses the whole strip rather than one contract, it is described as model free, meaning it does not depend on an option pricing model the way a single contract's implied volatility does.

The level moves with what option premiums cost. Demand for protection rises fastest when an underlying market falls sharply, so readings and prices have historically moved in opposite directions often enough for commentary to call such an index a fear gauge, a nickname that is loose: the calculation counts expected movement in both directions, and it is the pattern of demand for downside protection, rather than the arithmetic, that tilts it. Readings have also been strongly mean reverting historically, spending most of their time in a band and spiking briefly out of it.

Two things trip people repeatedly. The index is a calculation and not a portfolio, so there is nothing to hold: exposure exists only through listed futures and options written on it, and those price the expected level at their own expiry rather than the level today, which is why an instrument tracking them can fall over a period in which the index rose. And the published number is annualised, so it is not the movement expected over the window it describes; converting it to that window means dividing by the square root of the number of such windows in a year. How much information the level carries beyond recent realised volatility is genuinely disputed in the academic literature.