Partial Correlation Threshold Network Analysis of Malaysia Stock Market

Abstract

It is for years that Pearson cross correlation is used as a tool to monitor the relationship between a pair of equity returns. Incapability of cross correlation in distinguishing the role of a third stock in the correlation between any pair of stocks caused proposing partial correlation. This research first calculates the cross correlation of log return closing price of 1098 active Malaysia stocks in period of 246 trading days of 2016. A moving window of size five (five working days per week) is considered whereby the finial correlation value between any two stocks is the average of all calculated windows. Results are used to calculate the statistical measure of partial correlation to show how the correlation between any two stocks is affected by a third stock. The Partial Correlation Threshold Network (PCTN) of the obtained results is created where vertices are stocks and edges between a pair of nodes represents their partial influence. The PCTNs are investigated for different K threshold value of PCTN equation, in terms of number of Strongly Connected Components (SCC). Finally, the PCTN with maximum number of SCC is analyzed. Based on our observations, stocks belonged to Trading-Services group have more influence on other stocks.