Title: Ordinal Synchronization and Typical States in High-Frequency Digital Markets

Abstract: In this paper we study Algorithmic High-Frequency Financial Markets as dynamical networks. After an individual analysis of 24 stocks of the US market during a trading year of fully automated transactions by means of ordinal pattern series, we define an information-theoretic measure of pairwise synchronization for time series which allows us to study this subset of the US market as a dynamical network. We apply to the resulting network a couple of clustering algorithms in order to detect collective market states, characterized by their degree of centralized or descentralized synchronicity. This collective analysis has shown to reproduce, classify and explain the anomalous behavior previously observed at the individual level. We also find two whole coherent seasons of highly centralized and descentralized synchronicity, respectively. Finally, we model these states dynamics through a simple Markov model.