Volatility, irregularity, and predictable degree of accumulative return series.

Recently it was shown that financial time series are not completely random process but exhibit long-term or short-term dependences, which offer promises for predictability. However, we do not clearly understand the potential relationship between serial structure and predictability. This paper proposed a framework to magnify the correlations and regularities contained in financial time series through constructing accumulative return series. This method can help us distinguish the real world financial time series from random-walk process effectively by examining the change patterns of volatility, Hurst exponent, and approximate entropy. Furthermore, we have found that the predictable degree increases continually with the increasing length of accumulative return. Our results suggest that financial time series are predictable to some extent and approximate entropy is a good indicator to characterize the predictable degree of financial time series if we take the influence of their volatility into account.