Although I am completely aware they are simple functions I'd like to upload them because I noticed this distribution is not implemented by default and it could be useful for modelling logreturns distribution.
AV (2021). Generalized Error Distribution functions (https://www.mathworks.com/matlabcentral/fileexchange/57283-generalized-error-distribution-functions), MATLAB Central File Exchange. Retrieved .
The scale parameter refers to the possibility of "extending the distribution to a larger area":
the greater the scale parameter is, the more spread out the distribution will be.
You could try different alpha in GEDinv for a given beta and U~uni(0,1); and finally plot them and compare the differences.
Hope it will be useful.
Thanks man. I understand the beta concept. but what does scale mean? Is it the degree of freedom?
Sure, basically I used the same nomenclature of the wikipedia page:
alpha is the scale (it has to be a real and positive value), while beta is the shape value (positive and real).
Beta provides you the chance to obtain different distributions just modifying its value, for example if you use beta equals to 2 you obtain a Gaussian distribution (excess of Kurtosis = 0), beta equals to infinity gives you an uniform distribution and for beta minor than 2 you have "fatter" tails than the Gaussian and finallly for beta greater than 2 you have "tinner" tails compared to the Gaussian.
You could try to use GEDinv(p,alpha,beta) jointly with Monte Carlo simulation for p and than plot the results to have graphical evidences of what above exposed.
Hope I was clear enough, for any further requests do not hesitate to contact me.
Can you please guide me what is the beta and alpha component here for GED Inverse?
Find the treasures in MATLAB Central and discover how the community can help you!Start Hunting!