Title: Detecting long and short memory via spectral methods

Abstract: We study the properties of memory of a financial time series adopting two different methods of analysis, the detrended fluctuation analysis (DFA) and the analysis of the power spectrum (PSA). The methods are applied on three time series: one of high-frequency returns, one of shuffled returns and one of absolute values of returns. We prove that both DFA and PSA give results in line with those obtained with standard econometrics measures of correlation.