Market Graphs and Portfolio Optimization: A Maximum Clique Analysis

Abstract

The growing importance of data modeling tools for rendering structure to the complexities of financial markets has attracted interest in the market graph as a network of stock correlations. However, the sheer scale of information in market graphs presents a challenge for translating the model into critical applications such as macroeconomic analysis and portfolio investing. To extract meaningful and readily interpretable insights from the market graph, we turn to the maximum clique problem of identifying the largest cluster of completely-connected nodes in a graph. Over various decision parameters for market graph construction such as time, correlation, and financial metrics of risk and return, we provide a holistic but comprehensive survey of how the maximum clique structure can be leveraged to reflect prominent market dynamics and indicate target stocks for building optimized portfolios.