Verificação da memória longa persistente no mercado de bitcoins: uma análise do expoente de Hurst ao longo do tempo
Carregando...
Arquivos
Data
2015-01-21
Autores
Orientador(res)
Lora, Mayra Ivanoff
Métricas
Título da Revista
ISSN da Revista
Título de Volume
Resumo
One of the main points in the capital market study is the discussion of market efficiency theory, in which case differs in relation to the real price behavior of most assets. This work aims to analyze the behavior of the Bitcoin Price Index (BPI) during the period from July 2010 to September 2014. Initially the hypothesis of random walk will be tested for the BPI. Then, long period correlations will be observed in financial time series using the Hurst exponent (H) as an analytical tool, which was initially used to calculate correlations in natural phenomena and then its scope has achieved the financial area. The study calculates the H exponent by distinct methods highlighting the R/S and DFA analysis. For the valuation of the exponent varying in time, it is used a 90 days moving window dislocating from 10 to 10 days. The calculation of the H exponent at different scales analyzes, for each day, the value in the last 360, 180 and 90 days respectively. The results show that the BPI index presents persistent long memory in almost the whole period of study. Furthermore, the analysis at different scales indicates the possibility of predicting turbulent events in the index for the same period. Finally, it was possible to prove the hypothesis of fractal markets for the series of BPI returns.