EXTREME VALUE THEORY

[scorpiomanoj]

Hi David,

While I was attempting to generate EVT based VaR at my workplace, one of my colleagues pointed out that EVT VaR should always be higher than Historical simulation VaR. Though I was convinced I was not able to reason out the rationale. I kindly request you to clarify in this regard.

Thanks in anticipation,
Halan Manoj Kumar.

 

[David Harper CFA FRM]

Hi Manoj,

I don't strictly agree with the statement. EVT is an approach to parameterizing the tail: I don't see why a parameter (analytical) tail is necessarily heavier than an empirical (historical) tail.

Unless (maybe?) he/she refers to something like an ES EVT: because expected shortfall must be greater than VaR.

Thanks, David

 

[[email protected]]

Thanks for sharing your knowledge