Turns out that a couple of really smart economists just won the nobel prize for figuring this problem out.
Two economists (Engle, Granger) have been awarded the 2003 Nobel prize in economics for their pioneering work in developing statistical methods that make economic models more realistic and useful.
Engle's insight was to notice that the volatility of many financial or economic variables changes over time -- periods of great change are followed by calmer periods -- whereas earlier statistical models had assumed a constant rate of volatility.
See section 3 of the Nobel org's paper synopsis:
http://www.nobel.se/economics/laureates/2003/ecoadv.pdf
http://socialize.morningstar.com/NewSoc ... 1065740940
http://cbs.marketwatch.com/news/story.a ... eid=intuit
If someone wanted to condense and summarize this for the rest of us, or enhance their simulations, that would be great :D
Kramer