Chasing the Black Swan in cryptocurrency markets by modeling cascading dynamics in communication networks

The goal of this research is to improve work in predicting price fluctuations inside cryptocurrency markets. Our research is based on the concept of “Black Swan” events and cascading behaviours in social networks. On this foundation a cascade model has been constructed which analyzes the data of a Twitter communication network on whether the network structure is able to support a complete cascade of individual node-level panic (or hype) behavior. Therefore, unlike previous researches, which approached the underlying problem of predicting asset prices by implementing sophisticated AI models and trained them on historical market data, the proposed method in this article focuses on engineering a feature which can be used by those AI models to improve their predictive performances. While the employed AI model only provided directional results, the cascade model gives first indications for how to predict strong price fluctuations in the observed markets, and thus justifies further in-depth analysis.