Cryptocurrency market structure: connecting emotions and economics

TL;DR

This study explores the complex interrelations between prices and social sentiment in the cryptocurrency market, revealing significant correlations and causal influences, especially highlighting the role of minor currencies in the overall structure.

Contribution

It introduces innovative methodologies for analyzing large, non-linear systems and provides the first comprehensive analysis of the entire cryptocurrency market structure.

Findings

Prices are significantly correlated with sentiment.

Major cryptocurrencies have central roles in price networks but marginal in sentiment networks.

Minor currencies significantly influence the overall dependency and causality structure.

Abstract

We study the dependency and causality structure of the cryptocurrency market investigating collective movements of both prices and social sentiment related to almost two thousand cryptocurrencies traded during the first six months of 2018. This is the first study of the whole cryptocurrency market structure. It introduces several rigorous innovative methodologies applicable to this and to several other complex systems where a large number of variables interact in a non-linear way, which is a distinctive feature of the digital economy. The analysis of the dependency structure reveals that prices are significantly correlated with sentiment. The major, most capitalised cryptocurrencies, such as bitcoin, have a central role in the price correlation network but only a marginal role in the sentiment network and in the network describing the interactions between the two. The study of the causality structure reveals a causality network that is consistently related with the correlation structures and shows that both prices cause sentiment and sentiment cause prices across currencies with the latter being stronger in size but smaller in number of significative interactions. Overall our study uncovers a complex and rich structure of interrelations where prices and sentiment influence each other both instantaneously and with lead-lag causal relations. A major finding is that minor currencies, with small capitalisation, play a crucial role in shaping the overall dependency and causality structure. Despite the high level of noise and the short time-series we verified that these networks are significant with all links statistically validated and with a structural organisation consistently reproduced across all networks.