Bitcoin time series under the lens of information theory' quantifiers

AURELIO FERNÁNDEZ BARIVIERA; OSVALDO A. ROSSO; LUCIANO ZUNINO

San Pablo

Workshop; Econophysics Colloquium 2016; 2016

Many economic data are recorded as a sequence of measurements equally spaced in time. This kind of data, commonly referred as time series, are usually the starting point for economic analysis. In this line, information-theory-derived quantifiers can help to extract relevant information from financial time series. When studying dynamical systems, the discrimination of the presence of correlations in time series, emerges as one key task. In a recent paper, Bariviera et al. (2015) proposed the joint use of the Shannon entropy andthe Fisher Information Measure, as a proxy for informational efficiency. In another paper Bariviera et al. (2016) shows that the Complexity-Entropy Causality Plane constitutes a powerful graphical tool in order to discriminate stochastic and chaotic dynamics. Another element to take into account when studying time series is the probability density function estimation. We show that the use of the symbolic technique proposed by Bandt and Pompe (2002) is very useful in econophysics given its robustness to observational noise and absence of a priori assumptions. In this paper we extend our previous analysis, providing evidence that sampling frequency can uncover some additional characteristics of financial time series. We apply our technique to a time series of intraday bitcoin prices, in order to understand the behavior of this cryptocurrency.