What is Autoregressive Conditional Heteroskedasticity?

A nonlinear stochastic process, where the variance is time-varying, and a function of the past variance. ARCH processes have frequency distributions which have high peaks at the mean and fat-tails, much like fractal distributions. The ARCH model was inventedby Robert Engle. The Generalized ARCH (GARCH) model is the most widely used and was pioneered by Tim Bollerslev. See: Fractal Distributions.