Community Detection in Cryptocurrencies with Potential Applications to Portfolio Diversification

In this paper, the cross-correlations of cryptocurrency returns are analysed. The paper examines one years worth data for 146 cryptocurrencies from period January 1 2019 to December 31 2019. these firstly analysed by comparing eigenvalues and eigenvector components cross-correlation matrix C with Random Matrix Theory (RMT) assumptions. Results show that deviates assumptions indicating contains genuine information about correlations between different cryptocurrencies. From here, Louvain community detection method is applied as a clustering mechanism 15 groupings detected. Finally, PCA completed on standardised each clusters create portfolio investment. This selects which number high value coins when compared back against their market ranking in same year. interest assessing continuity initial results, also smaller dataset top 50 across three time periods T = 125 days, produces similar results. results obtained methods could be useful constructing optimally performing