0

My question is what would be the better method of VaR calculation among below two:, also any small code snippet will be great as a starting point for me.

1st method: I am trying to using a Generalized Pareto Distribution(GPD) there. I think R package POT or EVD might be of some help to fit my monthly historical return data to a GPD. Then using fExtremes package VaR might be calculated.

2nd Method : Another way is using PerformanceAnalytics package and trying calculate Modified Cornish-Fisher VaR.

LittleBobbyTables - Au Revoir
  • 32,008
  • 25
  • 109
  • 114
pmr
  • 998
  • 2
  • 13
  • 27
  • Unrelated comment: before fitting a GPD, you might want to check that a GPD is actually appropriate for your data. I've seen people use the GPD on long tails without thinking, when doing some exploration showed that the tail was actually better fit with an exponential. – Hong Ooi Aug 04 '13 at 08:28
  • @Hong Ooi, the GPD incorporates the exponential distribution as a special case (when the shape parameter of the GPD is equal to zero). So, the GPD fit can't be worse than the fit of the exponential distribution. – QuantIbex Aug 04 '13 at 17:54
  • @Mat, Better in terms of estimation accuracy, also is practically Modified Cornish-Fisher VaR is used in industry ? – pmr Aug 09 '13 at 05:23

0 Answers0