M.S. Candidate: Ersin Demirel
Program: Information Systems
Date: 18.01.2024 / 10:00
Place: A-212
Abstract: In the dynamic landscape of the stock market, accredited investors in the US are required to report quarterly holdings in their portfolios to the Securities and Exchange Commission (SEC) using Form 13F. Although these filings disclose holdings of each investor, it is challenging to extract valuable insights because of the complex structure of the market and the natural tendency of investors to keep their strategies discreet.
This thesis presents an innovative approach to process and enrich historical 13F filings into a dynamic bipartite graph of stocks and investors with rich edge and node attributes in order to shed light into investor behaviors in context of major crashes in the US stock markets. The analysis revealed that the clustering coefficient, significant sale and buy counts, late submitted filings and specific motif counts of the generated graph exhibit substantial changes during market crashes. These findings highlight the potential to use 13F filings and network methods for deeper understanding of market dynamics during downturns, and to be used as useful features for market monitoring systems.