Topological properties of a temporal recurrence network of financial time series

• Main Authors: Antenorcruz, Jude Maria V., Batac, Rene C.
• Format: text
• Published: Animo Repository 2024
• Subjects: Time-series analysis; System analysis; Stock exchanges; Finance and Financial Management; Statistics and Probability
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/13778
• Institution: De La Salle University

We create a temporally-directed network using the actual prices in a financial time series as the nodes and the temporal succession and the absolute closeness of their values as the criteria for link creation. Whenever a price is succeeded in time, it creates a record of the absolute price difference; subsequent occurrences of new price values will break the record when these new prices are closer to the original one, leading to new directed connections. This process is repeated until a recurrence, a repetition of an earlier price, is obtained. Interestingly, when applied to the time series of various financial markets, the distribution of the temporal separation between successive links shows robust power-law trends, revealing scale-free properties independent of the nature and type of markets. The technique presented here, which is based on complex network theory, may help reveal the underlying complexity involved in the generation of these time series and complement existing methods for the characterization of these highly dynamic, nonlinear phenomena.