José L. Acebal, Ana Paula P. Pereira, João Paulo O. Fernandes, Allbens A. Faria, Arthur Magalhães
Time series from financial assets are, in general, non stationary. They are exposed to action of many market agents driven by the information available. How much those time series are at random or determined by information is still a question that lacks explanation. After the time periods of relative stationarity are identified by statistical tests, the distribution of returns is used to generate a set of Lèvy flights. Then, q-gaussian and fractional derivative super and subdiffusive anomalous diffusion models are fitted by optimization in order to find the model parameters and to determine the regimen of anomalous diffusion associated to the generated flights. The result can shed light on risk assessment From information theory, further studies on exchanged entropy are conducted.