While computing R/S analysis, the range for each sub-period is calculated (R Ia), is then rescaled by the corresponding standard deviation (S Ia) depending on A contiguous sub-periods of length n. Here is my doubt, how can I compute the sub-periods?
There are several ways to choose subperiods in R/S analysis. The most common way is to choose non-overlapping subperiods covering whole data. So in the case of this method we should have time series of returns of length N. N ought to be a number having many dividers. According to Peters' book then we establish first length of subperiod (n>=10, n