Scaling analysis of trends using DFA

We study two aspects of the detrended fluctuation analysis (DFA) method, namely the scaling behavior of the leading terms of the best-fit polynomials and the detection of trends. We show analytically and numerically that the standard deviation of the leading terms of the best-fit polynomials used in DFA displays scaling behavior. Furthermore, we demonstrate that the distribution of these terms can be used to reveal the presence of trends in the data. We also argue that this distribution can be used as a sensitive tool for identifying weak trends.