Measures of association of local trends and networks of foreign exchange market in analysis of currency co-movement

In recent years it has become popular to represent the foreign exchange market as a correlation network using the Pearson correlation coefficient as a measure of co-movement of exchange rates. We show that the Pearson correlation of financial time series could be misleading in analyzing their co-movements. We propose representing the co-movement of exchange rates as a non-directed graph using the measure of local trends associations (LTA). Each node in the graph represents a currency, and an edge between nodes represents an existing high association between currencies. We present several methods for network summary visualization showing the highest associations between nodes. One method allows comparing graphs corresponding to different correlation and association measures. Another one is appropriate for comparing graphs using the same association measure. We present a dynamic analysis of association networks and the network of associations with a selected currency named a 'node of interest.' We show that the currency networks based on LTA are better explainable than networks based on Pearson correlation. LTA based relationships between currencies better reflect geographical, economic or political relationships between corresponding countries.