Quote from noregrets:
Cdcaveman, thank you--I wonder if my thinking has been off-base then. My impression was that--all other things being equal--a "low" volatility day was more likely to be followed by a "low" volatility day than by a "high" volatility day. Implied vol contains a huge amount of information about the future, but realized vol provides some important context as long as it is used with caution and understanding that shocks such as earnings, economic news, etc. are critical and obviously can't be derived from realized vol. Volatility regimes can exist, with perhaps an extreme example being the S&P going into periods of low volatility for years on end such as in the mid-00s.
Taleb in Dynamic Hedging, when describing ARCH/GARCH and its cousins, noted: "[ARCH] assumed that future volatility was linked to its past realizations, each realization having its own weight. Intuitively, it can be described as a process that can be forecast using past volatility data in a decreasing manner (as with the filtering technique). 'Volatility begets volatility,' as the saying goes. A volatile day is more likely to be followed by a volatile day and a quiet one by a quiet one...[GARCH, etc.] are getting closer to a seasoned option trader's opinion, which could be that past information about the price action impacts future volatility, but in some complex way. Subtle sets of rules in the trader's mind help him form his own opinion in a more potent way than these data-processing techniques." Taleb characterized this by noting that traders have "GARCH in their head." (p. 102)
