Pages

Sunday, October 6, 2013

Formulas references

Student's t-distribution
We start with a zero skew probability density (a more complicated version for skew-mixture will be shown in a subsequent post) of:

f(x) = Γ[(γ+1)/2] / [β*√(πγ)*Γ(γ/2)] * [1+1/γ*((x-α)/β)^2]^[-(γ+1)/2]

Gamma, Γ(z), is the positive integral of t^(z-1)*e^(-t)dt.

Average at γ<is undefined.  The shape parameter is γ.
Average at γ>1 is location parameter α

Variance at γ=1 is undefined and follows the Cauchy distribution
Variance at γ=2 is infinity
Variance at γ>2 is β^2*γ/(γ-2).  The scale parameter is β.

Skew at γ<3 is undefined
Skew at γ>3 is 0

Excess kurtosis (leptokurtosis) at γ<4 is undefined 
Excess kurtosis (leptokurtosis) at γ>4 is 6/(γ-4)
Excess kurtosis at γinfinity limits to 0

Tail distribution at finite γ follows the "power law" at power 1/(γ+1).  For example, how much greater is the event probability of a Cauchy versus at γ=3?
It is the square root of the probability, as a power of 1/(1+1)=1/2 is the root of a power of 1/(3+1)=1/4.


Check back on these pages regularly as we expand different formulae

No comments:

Post a Comment