Information Theory Quantifiers in Cryptocurrency Time Series Analysis

Area of Science:

• Quantitative Finance
• Complexity Science
• Data Science

Background:

• Cryptocurrency markets exhibit complex temporal dynamics.
• Understanding the randomness and chaos in financial time series is crucial for market analysis.
• Existing research often overlooks the interplay between project narratives and market behavior.

Purpose of the Study:

• To investigate the temporal evolution of cryptocurrency time series using information-theoretic measures.
• To differentiate between chaotic and stochastic behaviors in cryptocurrency price data.
• To assess the influence of white paper content on cryptocurrency market dynamics.

Main Methods:

• Applied information measures like complexity, entropy, and Fisher information to 176 daily cryptocurrency closing price time series.
• Utilized Complexity-Entropy Causality Plane (CECP) analysis to classify time series behavior.
• Employed Natural Language Processing (NLP) for white paper analysis and clustering, followed by time series dynamics comparison.

Main Results:

• Cryptocurrency time series under two years show chaotic behavior; series longer than two years exhibit stochastic behavior, often resembling colored noise (k between 0 and 2).
• NLP analysis revealed four distinct clusters based on white paper content.
• No significant correlation was found between white paper clusters and the time series dynamics.

Conclusions:

• Cryptocurrency market behavior transitions from chaotic to stochastic as data length increases.
• Project narratives, as reflected in white papers, do not appear to dictate short-to-medium term market dynamics.
• Investment strategies should prioritize real-time informational metrics over static white paper analysis.