Study of Equities using Signed Social Network

Abstract

Stock prices in the financial market are highly unpredictable; yet, there is usually some correlation between equities. These correlations can be positive or negative, causing equities to move in the same direction or in opposite directions. In this paper correlation analysis is performed on the Indian stock market to build a well-balanced portfolio. We have taken advantage of the fact that stocks have positive or negative correlation to create a signed social network of stocks using a modified winner-takes-all strategy [3]. We then used the FEC algorithm [9] to determine communities with positive intra-links and negative inter-links from this network. Our study revealed that various sub-sectors such as the pharmaceutical, IT, financial sector, and so on form communities within the network. We present a few different strategies to build portfolios by selecting companies from the determined communities that have only positive, only negative, or both negative and positive linkages. We use the Sharpe ratio[5] to assess the performance of these portfolios, and found t