GPD

[owais]

Hi All


Could you all please help me in calculating Tail Index of Stocks or modelling Tail distribution i am looking to model extreme risks say a non normal distribution using generalized pareto distribution in Excel VBA (any Code)

Any Help Suggetion

Thanks in Advance

 

[owais]

Any one here worked for calculating tail Index or code GPD generalized pareto distribution in VBA ???

 

[doug reich]

What are you looking for exactly? What do you have so far?

 

[owais]

I am looking to model that x% bad returns which cant be model with VaR i have done that with taking average of those bad days , i am assuming now a non normal distribution and looking for GPD to evaluate tail loss or Extreme loss for which i was asking advise how to do that in Excel VBA

I will glad if you help me


Many Thanks

 

[doug reich]

So is your approach monte carlo or something else?

 

[owais]

Not Monte carlo simulation i have done initial analysis using Historical simulation

 

[doug reich]

Now I'm confused -- historical simulation is model independent.

You're basically asking how to (a) fit a GPD to your historical returns (of the underlying), and (b) create a simulation of returns under that model, correct?

 

[owais]

Yes you are right now

I am so sorry for asking the question in that manner


In short I want to model Tail Loss or Extreme loss using GPD because i am assuming distribution which is non normal

 

[bob]

Haven't done any of this myself, but here's a paper that might come in handy. It describes several methods of extreme-value modeling, including among them generalized pareto, with examples and some notion of how you find parameters. For generalized pareto, it seems that setting your threshold for what constitutes an extreme return is the main challenge.

Good luck.

 

[jeff]

just use matlab

i've actually been trying to figure out how to do it in VBA also and then discovered that matlab actually does it automatically:

parmhat = gpfit(X)

http://www.mathworks.com/access/hel...vrvxZLG1iW6Q0c8Lw&sig2=te6HMW-Y3-wb9XcGTmkPxA