The RATS Software Forum

Dear Tom,

Is there a way to compute the skewness and kurtosis of the distribution of simulated returns? The table instruction does not provide these values. Please advise
Rachid

"Skewness" and "Kurtosis" are both in the index. You use the STATISTICS instruction, not TABLE. (TABLE only does a more limited set of statistics).

Many thank , I generated N= 5000 replications for vector series R1, yes, the Statistics instruction can give summary statistics for R1(1), R1(2)...., but I was wondering if you could get summary statistics for all R1. Thanks
Rachid

I'm not sure what you mean. You have R1(1),...,R1(5000) each of which is a series? Or you have R1(1),...,R1(n) each of which has 5000 entries? And what is it for which you want the statistics.

Thanks Tom, I have R1(1),...,R1(5000) each of which is a series, and would like to compute the overall skewness and kurtosis of R1 vector series .
Many thanks
Rachid

You would have to glue them into a single series. I'm not sure I understand what that's supposed to tell you. If you want to get a really good estimate of the process skewness and kurtosis just generate a series of length 1000000 and compute the statistics on that.

You originally referenced the TABLE instruction. TABLE does separate statistics on each series. How would that have helped if it did compute skewness and kurtosis?

Thanks for your inputs. my in-sample data has high kurtosis and relatively high negative skeweness, when I generate the returns of my model, I would like to know if I am generating the right numbers by looking at those two measures. I am currently doing it the you have suggested.
Thanks
Rachid