return distribution matching

jeffreylu · January 11th, 2007, 6:13 pm

suppose there is a sample return distribution and a group of different index return distributions. is it possible of finding the index return distributions which is cloest matching to the sample return distribution? any objective function/formula we can use as the criteria for the selection process?Many thanks in advance!

asd · January 11th, 2007, 7:11 pm

I would use a Q-Q plot , or try some statistical functions which compare the closeness of 2 distributions eg., kolmogorov smirnov test,etc.Hope it helps

jeffreylu · January 12th, 2007, 2:35 pm

Thanks a lot!But what is the difference between Kullback-Leibler Distance and kolmogorov smirnov test then?

mensa0 · January 14th, 2007, 8:13 am

The Kolmogorov-Smirnov test (KS-test) compares two sample distributions without regard to their underlying, possibly unknown, population distributions. The KullbackLeibler test compares a sample distribution with a known population distribution and is not useful for all distributions. The same is true of the Anderson-Darling test.Links:KSADKLMike

jeffreylu · January 15th, 2007, 9:43 am

Thanks a lot for the clear explanation!However, regarding the Kolmogorov-Smirnov test, how could tell from a group of index return distribution, which is the closest matching to the sample return distribution? What statistics I should use as a determinant criteria?