Skip to content

Latest commit

 

History

History
23 lines (18 loc) · 1.8 KB

README.md

File metadata and controls

23 lines (18 loc) · 1.8 KB

Multiplex Dependency Structure of S&P500 Stocks

alt text

Introduction

The aim of this work is to analyze dependencies and correlations between S&P500 stocks using a multiplex approach. To do so I considered different types of correlations (Pearson, Kendall and Spearman) between historical prices of stocks.

Build the Network

For each of the three correlation considered, I evaluated correlation matrices between historical log-returns of stocks. Then I filtered these matrices using the Planar Maximally Filtered Graph algorithm (PMFG). A python version of this algorithm that I wrote for this project is included in the notebooks folder. These preprocessing activity yields 3 filtered, weighted, undirected networks that, cosidered together, form a 3-layer multiplex network.

Multiplex Analysis

I considered a few multiplex metrics to analyze the multiplex stocks network:

  • Overlapping Degree, to check the global importance of nodes;
  • Multiplex Participation Coefficient (MPC), to analyze the distribution of the activity of a node among layers;
  • Multiplex Cartography;
  • Similarity between layers;

References