Exploring the Interconnectedness of Cryptocurrencies using Correlation\n Networks

Abstract

Correlation networks were used to detect characteristics which, although\nfixed over time, have an important influence on the evolution of prices over\ntime. Potentially important features were identified using the websites and\nwhitepapers of cryptocurrencies with the largest userbases. These were assessed\nusing two datasets to enhance robustness: one with fourteen cryptocurrencies\nbeginning from 9 November 2017, and a subset with nine cryptocurrencies\nstarting 9 September 2016, both ending 6 March 2018. Separately analysing the\nsubset of cryptocurrencies raised the number of data points from 115 to 537,\nand improved robustness to changes in relationships over time. Excluding USD\nTether, the results showed a positive association between different\ncryptocurrencies that was statistically significant. Robust, strong positive\nassociations were observed for six cryptocurrencies where one was a fork of the\nother; Bitcoin / Bitcoin Cash was an exception. There was evidence for the\nexistence of a group of cryptocurrencies particularly associated with Cardano,\nand a separate group correlated with Ethereum. The data was not consistent with\na token's functionality or creation mechanism being the dominant determinants\nof the evolution of prices over time but did suggest that factors other than\nspeculation contributed to the price.\n