Recurrence Network Analysis for Measuring Market Time-Series Structure
Summary
This article recasts a binary recurrence matrix as the adjacency matrix of an undirected network: observations become nodes, and recurrent pairs become edges, with self-links removed. It explains how network measures such as degree, clustering, transitivity, path length, betweenness, closeness, assortativity, and density can describe time-series structure beyond the line-pattern counts used in recurrence quantification analysis. It also extends the method to joint recurrence networks for analyzing synchronized recurrences in two series.
The article describes rolling-window tools and chart indicators for single-series and paired-series analysis, illustrating metrics with market-state interpretations such as clustering, fragmentation, and transitions between regimes. It is an analytical library tutorial, not a trading strategy, and reports no predictive or trading performance evidence. The author notes that some calculations, especially betweenness, scale cubically with window size; larger windows can therefore impose noticeable CPU costs. Metric interpretations are descriptive and do not by themselves establish a market signal.
Key ideas
- A recurrence matrix can be treated as a network adjacency matrix after removing its diagonal.
- Graph metrics capture clustering, connectivity, centrality, and degree relationships in recurrent states.
- Joint recurrence networks represent time points where two series recur simultaneously.
- Rolling indicators visualize selected network metrics for one or two market series.
- Network calculations can become computationally costly as the window size grows.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.