A Stock Market Filtering Model Based on Minimum Spanning Tree in Financial Networks

There have been several efforts in the literature to extract as much information as possible from the financial networks. Most of the research has been concerned about the hierarchical structures, clustering, topology and also the behavior of the market network; but not a notable work on the network filtration exists. This paper proposes a stock market filtering model using the correlation - based financial networks in which network nodes represent the potential stocks and network edges indicate the correlation coefficients of corresponding stock pairs. The model is capable of reducing the basic market size while keeping the diversification and risk - return expectations fairly constant. The novelty of this research is to develop a new market network filtering method which exploits Minimum Spanning Tree (MST) to reduce the number of network nodes (graph order) rather than the links (graph size). The proposed method chooses the nodes (stocks) based on dangling ends of the constructed MST. In order to verify our proposed model, we applied the model on data of three stock markets: New York Stock Exchange (NYSE), Germany Stock Exchange (DAX) and Toronto Stock Exchange (TSE). In conclusion, the numerical results showed that our proposed model can make a subset of the stock market in which its performance can imitate the whole market with a rather considerable reduction in size; as a result, we can have a diversified subset of the market compatible with that of the whole market. The performance of the model is confirmed by comparing the portfolio of the filtered market network with the whole market portfolio using the complement of Herfindahl Index as a measure of diversification.