Mis-Specification in the Estimation of the Expected Rescaled Adjusted Range Statistic: The Case Versus Peters
|Date of publication:||October 1996|
|Working paper number:||69|
Rescaled range analysis has in recent times gained in popularity as a means of identifying long memory effects in financial and economic time series data. Conclusions derived from the rescaled adjusted range statistic are conditional however upon the choice of an approptiate benchmark against which observed results can be compared. This paper provides an examination of various models of the expected value of the rescaled adjusted range statistic E(R*/sigma)_n. Two particular models will be cited, those of Anis and Lloyd (1976) and Peters (1994). As will be shown however, there exists significant inconsistencies in empirical results reported by Peters (1994), which when considered reveal Peters' specification of E(R*/sigma)_n should be rejected in favour of that derived by Anis and Lloyd.
|Paper:||Download (Format: PDF, Size: 730 Kb)|
|Comments:||Published as: Ellis, C., 2006, "The Mis-Specification of the Expected Rescaled Adjusted Range", Physica A: Statistical Mechanics and its Applications, 363(2), 469-476.|
Batten, J. A., Ellis, C. A. Fethertson, T. A., 2008, "Sample Period Selection and Long-Term Dependence: New Evidence from the Dow Jones Index", Chaos, Solitons & Fractals, 36(5), 1126-1140.
Ellis, C., 2007, "The Sampling Properties of Hurst Exponent Estimates", Physica A: Statistical Mechanics and its Applications, 375(1), 159-173.
Howe, J. S., Martin, D. W. and Wood Jr, B. G., 1999, "Much Ado About Nothing: Long-Term Memory in Pacific Rim Equity Markets", International Review of Financial Analysis, 8(2), 139-151.
Liu, Z., Zhang, H. and Dong, K., 2010, "A New Method to Determine the Periodicity of Time Series", Applied Mechanics and Materials, 26-28, 535-538.