This project employs graph theory to analyze financial portfolios. It involves creating different graph structures based on correlation, Euclidean distance, and statistical significance, and then analyzing these graphs using various centrality measures.
- MATLAB
- Data File:
projectdata_time_series.mat
- The dataset used in this project contains time series data for financial portfolio analysis.
This project utilizes the dataset provided by Kozak et al. If you use the data or the analysis provided here, please include the following citation:
Kozak, S., Nagel, S., Santosh, S., & Daniel, K. (2018). Discussion of: Shrinking the Cross Section.
Clone this repository or download the script files and the required data file projectdata_time_series.mat
.
- Objective: Generate graphs based on correlation, Euclidean distance, and statistical significance.
- Output: Three graphs G1, G2, G3 representing different relationships between portfolios.
- Objective: Compute degree, eigenvector, and PageRank centralities for each graph.
- Output: Centrality values for each node in the graphs.
- Objective: Visualize the centrality measures for each graph.
- Visualization: Bar graphs showing the centrality distribution of portfolios in each graph.
- Objective: Rank and visualize portfolios based on PageRank centrality.
- Visualization: Bar graphs displaying ranked centrality scores, highlighting the most influential portfolios.
This analysis can be used for risk assessment, portfolio management, and investment strategy development by identifying key portfolios and understanding portfolio interactions.
For queries and contributions, please contact Abdulmalik Abdukayumov (abdumalik.abdukayumov@duke.edu) or Defne Circi.