Stable distribution explained

In probability theory, a distribution is said to be stable if a linear combination of two independent random variables with this distribution has the same distribution, up to location and scale parameters. A random variable is said to be stable if its distribution is stable. The stable distribution family is also sometimes referred to as the Lévy alpha-stable distribution, after Paul Lévy, the first mathematician to have studied it.^{[1] [2]}

Of the four parameters defining the family, most attention has been focused on the stability parameter,

\alpha (see panel). Stable distributions have 0 < \alpha \leq 2 , with the upper bound corresponding to the normal distribution, and \alpha =1 to the Cauchy distribution. The distributions have undefined variance for \alpha < 2 , and undefined mean for \alpha \leq 1 . The importance of stable probability distributions is that they are "attractors" for properly normed sums of independent and identically distributed (iid) random variables. The normal distribution defines a family of stable distributions. By the classical central limit theorem the properly normed sum of a set of random variables, each with finite variance, will tend toward a normal distribution as the number of variables increases. Without the finite variance assumption, the limit may be a stable distribution that is not normal. Mandelbrot referred to such distributions as "stable Paretian distributions",^{[3] [4] [5]} after Vilfredo Pareto. In particular, he referred to those maximally skewed in the positive direction with 1 < \alpha < 2 as "Pareto–Lévy distributions",^[1] which he regarded as better descriptions of stock and commodity prices than normal distributions.^[6]

Definition

A non-degenerate distribution is a stable distribution if it satisfies the following property:

Since the normal distribution, the Cauchy distribution, and the Lévy distribution all have the above property, it follows that they are special cases of stable distributions.

Such distributions form a four-parameter family of continuous probability distributions parametrized by location and scale parameters μ and c, respectively, and two shape parameters

\beta and \alpha , roughly corresponding to measures of asymmetry and concentration, respectively (see the figures).\varphi(t) of any probability distribution is the Fourier transform of its probability density function f(x) . The density function is therefore the inverse Fourier transform of the characteristic function:^[7] $$\backslash varphi(t)\; =\; \backslash int\_^\backslash infty\; f(x)e^\backslash ,dx.$$

Although the probability density function for a general stable distribution cannot be written analytically, the general characteristic function can be expressed analytically. A random variable X is called stable if its characteristic function can be written as^[8] $$\backslash varphi(t;\; \backslash alpha,\; \backslash beta,\; c,\; \backslash mu)\; =\; \backslash exp\; \backslash left\; (i\; t\; \backslash mu\; -\; |c\; t|^\backslash alpha\; \backslash left\; (1\; -\; i\; \backslash beta\; \backslash sgn(t)\; \backslash Phi\; \backslash right)\; \backslash right)$$where is just the sign of andμ ∈ R is a shift parameter,

\beta \in [ -1 , 1 ] , called the skewness parameter, is a measure of asymmetry. Notice that in this context the usual skewness is not well defined, as for \alpha < 2 the distribution does not admit 2nd or higher moments, and the usual skewness definition is the 3rd central moment.

The reason this gives a stable distribution is that the characteristic function for the sum of two independent random variables equals the product of the two corresponding characteristic functions. Adding two random variables from a stable distribution gives something with the same values of

\alpha and \beta , but possibly different values of μ and c.

Not every function is the characteristic function of a legitimate probability distribution (that is, one whose cumulative distribution function is real and goes from 0 to 1 without decreasing), but the characteristic functions given above will be legitimate so long as the parameters are in their ranges. The value of the characteristic function at some value t is the complex conjugate of its value at −t as it should be so that the probability distribution function will be real.

In the simplest case

\beta = 0 , the characteristic function is just a stretched exponential function; the distribution is symmetric about μ and is referred to as a (Lévy) symmetric alpha-stable distribution, often abbreviated SαS.

When

\alpha < 1 and \beta = 1 , the distribution is supported on [''μ'', ∞). The parameter ''c'' > 0 is a scale factor which is a measure of the width of the distribution while <math>\alpha</math> is the exponent or index of the distribution and specifies the asymptotic behavior of the distribution. === Parametrizations === The parametrization of stable distributions is not unique. Nolan <ref name="Nolan2020" /> tabulates 11 parametrizations seen in the literature and gives conversion formulas. The two most commonly used parametrizations are the one above (Nolan's "1") and the one immediately below (Nolan's "0"). The parametrization above is easiest to use for theoretical work, but its probability density is not continuous in the parameters at <math>\alpha =1</math>.<ref name="Nolan 1997"/> A continuous parametrization, better for numerical work, is<ref name=":0" /> <math display="block"> \varphi(t; \alpha, \beta, \gamma, \delta) = \exp \left (i t \delta - |\gamma t|^\alpha \left (1 - i \beta \sgn(t) \Phi \right) \right) </math> where: <math display="block"> \Phi = \begin{cases} \left (|\gamma t|^{1 - \alpha} - 1 \right) \tan \left (\tfrac{\pi \alpha}{2} \right) & \alpha \neq 1 \\ - \frac{2}{\pi} \log|\gamma t| & \alpha = 1 \end{cases} </math> The ranges of <math>\alpha</math> and <math>\beta</math> are the same as before, ''γ'' (like ''c'') should be positive, and ''δ'' (like ''μ'') should be real. In either parametrization one can make a linear transformation of the random variable to get a random variable whose density is <math> f(y; \alpha, \beta, 1, 0) </math>. In the first parametrization, this is done by defining the new variable: <math display="block"> y = \begin{cases} \frac{x - \mu}\gamma & \alpha \neq 1 \\ \frac{x - \mu}\gamma - \beta\frac 2\pi\ln\gamma & \alpha = 1 \end{cases} </math> For the second parametrization, simply use <math display="block"> y = \frac{x-\delta}\gamma </math> independent of <math>\alpha</math>. In the first parametrization, if the mean exists (that is, <math>\alpha > 1</math>) then it is equal to ''μ'', whereas in the second parametrization when the mean exists it is equal to <math> \delta - \beta \gamma \tan \left (\tfrac{\pi\alpha}{2} \right).</math> ===The distribution=== A stable distribution is therefore specified by the above four parameters. It can be shown that any non-degenerate stable distribution has a smooth (infinitely differentiable) density function.<ref name=":0" /> If <math> f(x; \alpha, \beta, c, \mu) </math> denotes the density of ''X'' and ''Y'' is the sum of independent copies of ''X'': <math display="block"> Y = \sum_{i = 1}^N k_i (X_i - \mu)</math> then ''Y'' has the density <math> \tfrac{1}{s} f(y / s; \alpha, \beta, c, 0) </math> with <math display="block"> s = \left(\sum_{i = 1}^N |k_i|^\alpha \right)^{\frac{1}{\alpha}} </math> The asymptotic behavior is described, for <math>\alpha < 2</math>, by:<ref name=":0" /> <math display="block"> f(x) \sim \frac{1}{|x|^{1 + \alpha}} \left (c^\alpha (1 + \sgn(x) \beta) \sin \left (\frac{\pi \alpha}{2} \right) \frac{\Gamma(\alpha + 1) }{\pi} \right) </math> where Γ is the [[Gamma function]] (except that when \alpha \geq 1 and \beta = \pm 1 , the tail does not vanish to the left or right, resp., of μ, although the above expression is 0). This "heavy tail" behavior causes the variance of stable distributions to be infinite for all \alpha < 2 . This property is illustrated in the log–log plots below.

When

\alpha = 2 , the distribution is Gaussian (see below), with tails asymptotic to exp(−x²/4c²)/(2c√π).

One-sided stable distribution and stable count distribution

When

\alpha < 1 and \beta = 1 , the distribution is supported on [''μ'', ∞). This family is called '''one-sided stable distribution'''.<ref name="PhysRevLett 1007">{{Cite journal| last1=Penson|first1=K. A.| last2=Górska|first2=K.| date=2010-11-17| title=Exact and Explicit Probability Densities for One-Sided Lévy Stable Distributions| journal=Physical Review Letters| volume=105 | issue=21 | pages=210604 | doi=10.1103/PhysRevLett.105.210604| pmid=21231282| arxiv=1007.0193| bibcode=2010PhRvL.105u0604P | s2cid=27497684}}</ref> Its standard distribution (μ=0) is defined as :<math>L_\alpha(x) = f{\left(x;\alpha,1,\cos\left(\frac{\alpha\pi}{2}\right)^{1/\alpha},0\right)}</math>, where <math>\alpha < 1</math>. Let <math>q = \exp(-i\alpha\pi/2)</math>, its characteristic function is <math> \varphi(t;\alpha) = \exp\left (- q|t|^\alpha \right) </math>. Thus the integral form of its PDF is (note: <math>\operatorname{Im}(q)<0</math>) <math display="block"> \begin{align} L_\alpha(x) & = \frac{1}{\pi}\Re\left[ \int_{-\infty}^\infty e^{itx}e^{-q|t|^\alpha}\,dt\right]\\ & = \frac \int_0^\infty e^ \sin(tx)\sin(-\operatorname(q)\,t^\alpha) \,dt, \text\\ & = \frac \int_0^\infty e^ \cos(tx)\cos(\operatorname(q)\,t^\alpha) \,dt .\end

The double-sine integral is more effective for very small

x .

Consider the Lévy sum $Y\; =\; \backslash sum\_^N\; X\_i$ where $X\_i\; \backslash sim\; L\_\backslash alpha(x)$, then Y has the density $\backslash frac\; L\_\backslash alpha$ where $\backslash nu\; =\; N^$. Set $x\; =\; 1$ to arrive at the stable count distribution. Its standard distribution is defined as

ak{N} _{\alpha(\nu) = (\alpha)/(\Gamma\left( (1)/(\right) \alpha))} (1)/(\nu) L _{\alpha{\left( (1)/(\nu)} \right)} , where \nu > 0 and \alpha < 1 .

The stable count distribution is the conjugate prior of the one-sided stable distribution. Its location-scale family is defined as

ak{N} _{\alpha(\nu;\nu} _{0,\theta)} = (\alpha)/(\Gamma( (1)/() \alpha)) (1)/(\nu - \nu _{0}) L _{\alpha{\left( (\theta)/(\nu - \nu _{0})} \right)} , where \nu > \nu _{0} , \theta > 0 , and \alpha < 1 .

It is also a one-sided distribution supported on

[\nu _{0,infty)} . The location parameter \nu _{0} is the cut-off location, while \theta defines its scale.

When $\backslash alpha\; =\; \backslash frac$, $L\_(x)$ is the Lévy distribution which is an inverse gamma distribution. Thus

ak{N} _{(1)/(2)} (\nu; \nu _{0,} \theta) is a shifted gamma distribution of shape 3/2 and scale 4 \theta ,ak{N} _{(1)/(2)} (\nu;\nu _{0,\theta)} = (1)/(4 \sqrt{\pi) \theta ^{3/2} }(\nu-\nu_0)^ e^, where \nu > \nu _{0} , \theta > 0 .

Its mean is

\nu _{0} + 6 \theta and its standard deviation is \sqrt{ 24 }\theta . It is hypothesized that VIX is distributed like $\backslash mathfrak\_(\backslash nu;\backslash nu\_0,\backslash theta)$ with \nu _{0} = 10 . 4 and \theta = 1 . 6 (See Section 7 of). Thus the stable count distribution is the first-order marginal distribution of a volatility process. In this context, \nu _{0} is called the "floor volatility".

Another approach to derive the stable count distribution is to use the Laplace transform of the one-sided stable distribution, (Section 2.4 of)

\int ^{inf ty} _{0} e ^{-z} L _{\alpha(x)} dx = e ^{- z ^{\alpha}} , where \alpha< 1 .

Let

x = 1 / \nu , and one can decompose the integral on the left hand side as a product distribution of a standard Laplace distribution and a standard stable count distribution,f\int ^{inf ty} _{0} (1)/(\nu) \left ( (1)/(2) e ^{- (|z|)/(\nu)} \right ) \left ( (\alpha)/(\Gamma( (1)/() \alpha)) (1)/(\nu) L _{\alpha{\left( (1)/(\nu)} \right)} \right ) d \nu = (1)/(2) (\alpha)/(\Gamma( (1)/() \alpha)) e ^{- |z| ^{\alpha}} , where \alpha< 1 .

This is called the "lambda decomposition" (See Section 4 of) since the right hand side was named as "symmetric lambda distribution" in Lihn's former works. However, it has several more popular names such as "exponential power distribution", or the "generalized error/normal distribution", often referred to when

\alpha > 1 .

The n-th moment of

ak{N} _{\alpha(\nu)} is the - (n + 1 ) -th moment of L _{\alpha(x)} , and all positive moments are finite.

Properties

• All stable distributions are infinitely divisible.
• With the exception of the normal distribution (

\alpha = 2 ), stable distributions are leptokurtotic and heavy-tailed distributions.

• Closure under convolution

Stable distributions are closed under convolution for a fixed value of

\alpha . Since convolution is equivalent to multiplication of the Fourier-transformed function, it follows that the product of two stable characteristic functions with the same \alpha will yield another such characteristic function. The product of two stable characteristic functions is given by:$$\backslash exp\backslash left\; (it\backslash mu\_1+it\backslash mu\_2\; -\; |c\_1\; t|^\backslash alpha\; -\; |c\_2\; t|^\backslash alpha\; +i\backslash beta\_1|c\_1\; t|^\backslash alpha\backslash sgn(t)\backslash Phi\; +\; i\backslash beta\_2|c\_2\; t|^\backslash alpha\backslash sgn(t)\backslash Phi\; \backslash right)$$

Since is not a function of the μ, c or

\beta variables it follows that these parameters for the convolved function are given by:

In each case, it can be shown that the resulting parameters lie within the required intervals for a stable distribution.

The Generalized Central Limit Theorem

The Generalized Central Limit Theorem (GCLT) was an effort of multiple mathematicians (Berstein, Lindeberg, Lévy, Feller, Kolmogorov, and others) over the period from 1920 to 1937.^[9] The first published complete proof (in French) of the GCLT was in 1937 by Paul Lévy.^[10] An English language version of the complete proof of the GCLT is available in the translation of Gnedenko and Kolmogorov's 1954 book.^[11]

The statement of the GLCT is as follows:^[12]

A non-degenerate random variable Z is α-stable for some 0 < α ≤ 2 if and only if there is an independent, identically distributed sequence of random variables X₁, X₂, X₃, ... and constants a_n > 0, b_n ∈ ℝ with

a_n (X₁ + ... + X_n) - b_n → Z.

Here → means the sequence of random variable sums converges in distribution; i.e., the corresponding distributions satisfy F_n(y) → F(y) at all continuity points of F.

In other words, if sums of independent, identically distributed random variables converge in distribution to some Z, then Z must be a stable distribution.

Special cases

There is no general analytic solution for the form of f(x). There are, however three special cases which can be expressed in terms of elementary functions as can be seen by inspection of the characteristic function:^[13]

• For

\alpha = 2 the distribution reduces to a Gaussian distribution with variance σ² = 2c² and mean μ; the skewness parameter \beta has no effect.

• For

\alpha = 1 and \beta = 0 the distribution reduces to a Cauchy distribution with scale parameter c and shift parameter μ.

• For

\alpha = 1 / 2 and \beta = 1 the distribution reduces to a Lévy distribution with scale parameter c and shift parameter μ.

Note that the above three distributions are also connected, in the following way: A standard Cauchy random variable can be viewed as a mixture of Gaussian random variables (all with mean zero), with the variance being drawn from a standard Lévy distribution. And in fact this is a special case of a more general theorem (See p. 59 of ^[14]) which allows any symmetric alpha-stable distribution to be viewed in this way (with the alpha parameter of the mixture distribution equal to twice the alpha parameter of the mixing distribution—and the beta parameter of the mixing distribution always equal to one).

A general closed form expression for stable PDFs with rational values of

\alpha is available in terms of Meijer G-functions.^[15] Fox H-Functions can also be used to express the stable probability density functions. For simple rational numbers, the closed form expression is often in terms of less complicated special functions. Several closed form expressions having rather simple expressions in terms of special functions are available. In the table below, PDFs expressible by elementary functions are indicated by an E and those that are expressible by special functions are indicated by an s.

		\alpha
		1/3	1/2	2/3	1	4/3	3/2	2
\beta	0	s	s	s	E	s	s	E
	1	s	E	s	L		s

Some of the special cases are known by particular names:

• For

\alpha = 1 and \beta = 1 , the distribution is a Landau distribution (L) which has a specific usage in physics under this name.

• For

\alpha = 3 / 2 and \beta = 0 the distribution reduces to a Holtsmark distribution with scale parameter c and shift parameter μ.

Also, in the limit as c approaches zero or as α approaches zero the distribution will approach a Dirac delta function .

Series representation

The stable distribution can be restated as the real part of a simpler integral:^[16] $$f(x;\backslash alpha,\backslash beta,c,\backslash mu)=\backslash frac\backslash Re\backslash left[\backslash int\_0^\backslash infty\; e^\{it(x-\backslash mu)\}e^\{-(ct)^\backslash alpha(1-i\backslash beta\backslash Phi)\}\backslash ,dt\backslash right].$$

Expressing the second exponential as a Taylor series, this leads to:$$f(x;\backslash alpha,\backslash beta,c,\backslash mu)=\backslash frac\backslash Re\backslash left[\backslash int\_0^\backslash infty\; e^\{it(x-\backslash mu)\}\backslash sum\_\{n=0\}^\backslash infty\backslash frac\{(-qt^\backslash alpha)^n\}\{n!\}\backslash ,dt\backslash right]$$where

q = c ^{\alpha(1-i\beta\Phi)} . Reversing the order of integration and summation, and carrying out the integration yields:$$f(x;\backslash alpha,\backslash beta,c,\backslash mu)=\backslash frac\backslash Re\backslash left[\backslash sum\_\{n=1\}^\backslash infty\backslash frac\{(-q)^n\}\{n!\}\backslash left(\backslash frac\{i\}\{x-\backslash mu\}\backslash right)^\{\backslash alpha\; n+1\}\backslash Gamma(\backslash alpha\; n+1)\backslash right]$$which will be valid for x ≠ μ and will converge for appropriate values of the parameters. (Note that the n = 0 term which yields a delta function in x − μ has therefore been dropped.) Expressing the first exponential as a series will yield another series in positive powers of x − μ which is generally less useful.

For one-sided stable distribution, the above series expansion needs to be modified, since

q = \exp( - i\alpha\pi/ 2 ) and q i ^{\alpha} =1 . There is no real part to sum. Instead, the integral of the characteristic function should be carried out on the negative axis, which yields:^[17]

Notes and References

1. B. . Mandelbrot . The Pareto–Lévy Law and the Distribution of Income . International Economic Review . 1 . 2 . 1960 . 79–106 . 10.2307/2525289 . 2525289 .
2. Book: Lévy, Paul . Calcul des probabilités . Paris . Gauthier-Villars . 1925 . 1417531 .
3. B. . Mandelbrot . Stable Paretian Random Functions and the Multiplicative Variation of Income . Econometrica . 29 . 4 . 517–543 . 1961 . 10.2307/1911802 . 1911802 .
4. B. . Mandelbrot . The Variation of Certain Speculative Prices . The Journal of Business . 36 . 4 . 394–419 . 1963 . 10.1086/294632 . 2350970 .
5. Eugene F. . Fama . Mandelbrot and the Stable Paretian Hypothesis . The Journal of Business . 36 . 4 . 420–429 . 1963 . 10.1086/294633 . 2350971 .
6. Mandelbrot . B. . New methods in statistical economics . . 71 . 5 . 421–440 . 1963 . 10.1086/258792 . 53004476 .
7. Web site: Stable Distributions. Siegrist. Kyle. www.randomservices.org. en. 2018-10-18.
8. Book: Voit, Johannes. The Statistical Mechanics of Financial Markets – Springer . 10.1007/b137351. W. Thirring. H. Grosse . W. Beiglböck. R. Balian. Springer. 2005 . Texts and Monographs in Physics. 978-3-540-26285-5.
9. Le Cam . L. . The Central Limit Theorem around 1935 . Statistical Science . February 1986 . 1 . 1 . 78–91 . 2245503 .
10. Book: Lévy . Paul . Theorie de l'addition des variables aleatoires [Combination theory of unpredictable variables] . 1937 . Gauthier-Villars . Paris.
11. Book: Gnedenko . Boris Vladimirovich . Kologorov . Andreĭ Nikolaevich . Doob . Joseph L. . Hsu . Pao-Lu . Limit distributions for sums of independent random variables . 1968 . Addison-wesley . Reading, MA.
12. Book: Nolan . John P. . Univariate stable distributions, Models for Heavy Tailed Data . Springer Series in Operations Research and Financial Engineering . 2020 . Springer . Switzerland . 10.1007/978-3-030-52915-4 . 978-3-030-52914-7 . 226648987 .
13. Book: Stable Non-Gaussian Random Processes: Stochastic Models with Infinite Variance. Samorodnitsky. G.. CRC Press. 1994. 9780412051715. Taqqu. M.S..
14. Book: Lee, Wai Ha. Continuous and discrete properties of stochastic processes. PhD thesis, University of Nottingham. 2010.
15. On Representation of Densities of Stable Laws by Special Functions. Theory of Probability and Its Applications. 1995. 0040-585X. 354–362. 39. 2. 10.1137/1139025. V.. Zolotarev.
16. Theory of the pressure broadening and shift of spectral lines. Advances in Physics. 1981. 0001-8732. 367–474 . 30. 3. 10.1080/00018738100101467. G.. Peach. 1981AdPhy..30..367P.
17. Pollard. Howard. 1946. Representation of e^ As a Laplace Integral|url=https://projecteuclid.org/euclid.bams/1183509728|journal=Bull. Amer. Math. Soc. |volume=52|pages=908|doi=10.1090/S0002-9904-1946-08672-3|via=|doi-access=free}.