Comparison of two types of topological networks for the foreign exchange market: one based on correlation coefficients and the other on the concept of causality

Topological networks make it possible to recognize structural properties of the currency market. Such networks can be constructed on the basis of the values of correlation coefficients between currency pairs, and the popular minimum spanning tree (MST) algorithm allows an understanding of significant relationships on the market. An alternative measure of distance, based on the concept of causality for time series, makes it possible to measure not only the strength of relationships between currency pairs but also the directionality of these relationships. There is even an equivalent of MST on a directed graph ? minimum-cost arborescence (MCA). The purpose of this study is to compare correlation and causality networks built for the foreign exchange market. The networks were constructed in a stepwise manner for the most important world currencies in the period from 3rd Jan. 2020 to 18th Oct. 2024. The comparison was carried out using certain topological characteristics of the networks, such as density, average distance, diameter, centralization index and degrees of vertices. The study details the properties of both approaches.