Topological Analysis of the Transfer Entropy Network of Cryptocurrencies During the Uptrend and Downtrend Periods

This research aims to understand the cryptocurrency market behaviour between downward and upward trends through the networks that were built from transfer entropy (TE). Calculations were performed on two datasets consisting of time series of 40 cryptocurrency prices: (1) the ‘upward trend’ dataset with an obvious increase of the Bitcoin price and (2) the ‘downward trend’ dataset with a remarkable decrease of the Bitcoin price. Then, information flow networks were created by TE value from each pair of cryptocurrencies. Lastly, we compared the network topology of upward and downward trends of the market to find how differently the market behave. We found that in the dataset with 5-minute interval, the amount of significant TE is more than that in 1-hour data, but the TE values in the 1-hour data was greater than the those in 5-minute data. Our result also showed that the cryptocurrency market was very sensitive, and cryptocurrency holders were more likely to trade and behave according to the information they receive during a downward trend rather than an upward trend and the result prefer us to use information in bigger time step (e.g., 1 Hour > 5 minute) to reduce the risk when decide to invest in cryptocurrency by using history information of another cryptocurrency that have causal relation with.