Stats › Volatility and VIX › Volatility skew
Implied volatility skew
If you take two options on the same index, expiring on the same day, one the same distance below the market and one the same distance above, the standard textbook says they should carry the same implied volatility. They don't. The downside one is more expensive, consistently, and has been for nearly forty years. That gap is called skew.
What I find remarkable about it is that it has a start date. Cboe index options launched in 1983 and for the first four years there was almost no observable skew at all. Then came 19 October 1987, the S&P 500 fell 20.47% in a session, and the skew appeared and never left. Cboe's own words for it are that the day "marked the end of history in option markets and the dawn of a new era, an age defined by the existence of skew". This page covers what skew is in plain terms, how it began, what Cboe's SKEW Index actually measures, and the evidence on whether it predicts anything, which is thinner than its reputation.
Fair warning: this is the most technical page in the library and I've tried to explain each term once rather than assume it. It's also the page where I've had to leave things out. There's no current SKEW level here and no record high, because the sources contradicted each other badly enough that publishing a number would have been guessing. The notes set out exactly what failed and why, including one important paper I couldn't retrieve, which tilts the balance of this page toward the sceptical side. You should know that before you read the rest.
Before October 1987 equity index options with equal distance above and below the market traded at roughly the same implied volatility. Cboe's own account is blunt: Black Monday "marked the end of history in option markets and the dawn of a new era, an age defined by the existence of skew". Since then downside puts have been persistently more expensive. Cboe's SKEW Index puts a number on it, with a stated historical range of 100 to 150 and an average around 115, where 100 means a normal distribution. The evidence that it predicts anything is weak: Cboe makes no predictive claim, and the published academic test found no statistically significant relationship between SKEW and realised skewness.
Members only
The rest of this page is for members
Below this point there are 8 sections, 1 chart and 4 named sources, roughly 2000 words of it. Every figure carries the source it came from and the date the data is from.
You can keep browsing every statistic in the library for free. The intro and the summary are always open.