Abstract:
The cryptocurrency market is one of the fastest expanding markets in the financial
world, especially for these last few years where it noticed a big entry of investors.
Cryptocurrencies are better known to be independent of any regulations, freeing
themselves from being under the authority of central banks, thus, it is acknowledged
that they are 100% decentralized. They mostly have a role of a store of value, even
though sometimes they can be considered as a medium of exchange too. This study
aims to forecast Bitcoin’s prices, to better understand its returns. The approach used
in forecasting is Auto-Regressive Integrated Moving Average (ARIMA). ACF,
PACF, and ADF tests were performed for data reliability, and to test the potential
trend, seasonality, and stationarity detections. Because the data was non-stationary,
the first differencing series were used after they became stationary.
Even the first lag suggested for the ARIMA order was (1,1,0), for a better
understanding of the study, and better rigor, 2 more lags as ARIMA (1,0,0) and
ARIMA (2,1,0) were implemented too. However, the test results revealed that the
preferred model to fit is (1,1,0).
