Cryptocurrencies and Decentralized Finance (DeFi) currently represent a fast growing trend in finance, which enables financial services on public blockchains. In contrast to traditional financial markets, ruled by well established corporations, DeFi is completely transparent, as it keeps publicly available records of all transactions that occur in the network. This availability of the data represents an opportunity to analyze and understand the market from the point of view of the complexity that emerges from the interactions among actors (users, bots, and companies) operating in the embedded market. In this paper, we focus on Ethereum to show that the underlying transaction network bears further and useful information to forecast the evolution of the market. We aim to separate the non-redundant effects of the blockchain transaction network from technical analysis and social media trends in the future price of the Ethereum native cryptocurrency. To this end, we build two machine learning models to predict the future trend of the price time series. The first model, serving as a base, considers the set of most relevant features according to the current scientific literature-including technical analysis and social media trends. The second model considers the features of the base model, incorporating the network properties computed from the transaction network. We find that the second model outperforms the base model and can anticipate 46% more rises in the price than the base model and 19% more falls. Thus, we conclude that new indicators based on network properties provide valuable information to forecast the future direction of the market that cannot be explained neither by technical analysis nor by social media trends alone. Hence, our results represent an important first step toward the definition of a new family of DeFi market indicators based on the complexity of the underlying transaction network.
Grande et al. (Fri,) studied this question.