Meaning
Probability models that retain their shape under addition define a broad family of continuous random variables. An alpha stable distribution is defined by four parameters that govern its location, scale, skewness, and tail heaviness. These mathematical models are particularly useful for representing variables that exhibit extreme fluctuations and heavy tails.
Dispersion Shape
Probability theory relies on the central limit theorem to explain why averages of many independent variables converge to specific patterns. While the classical theorem leads to the Gaussian distribution, the generalized version dictates that the limit must be an alpha stable distribution when the underlying variables have infinite variance. The shape parameter, often designated by the Greek letter alpha, lies between zero and two.
Lower values of this parameter generate higher peaks and thicker tails, which alters the speed of convergence in empirical studies.
Asymptotic Behavior
Extreme events in finance and network traffic often exceed the predictions of standard normal models. An alpha stable distribution provides a framework where the probability of large deviations decreases slowly. This slower decay creates a heavy tail that better matches actual observations of market returns and transaction spikes.
Parameter Bound
The estimation of these variables requires special numerical methods because most of them lack closed-form density functions. Apart from a few specific cases, practitioners must use characteristic functions to analyze the alpha stable distribution. This complexity limits its direct application to situations where tail behavior is the primary interest.