The standard complement of statistical techniques used to identify predictable market structure is only capable of identifying regular periodic cycles and assumes that the data are independent and identically distributed (i.i.d.). Yet, financial returns data are not independent and cycles are most probably not periodic. Rescaled range analysis is a nonparametric technique that can distinguish the average cycle length of irregular cycles. Using Australian stock market data, this paper finds evidence of long memory in the returns generating process and non-periodic cycles of approximately 3, 6 and 12 years in average duration.
|Number of pages||14|
|Publication status||Published - 2001|