Leiden Community Detection (IAQF Problem 2025)

The IAQF Student Problem 2025 details the growing concentration of equity markets driven by the Magnificent Seven, which collectively account for a disproportionate share of index returns, risk, and volatility transmission in major benchmarks such as the S&P 500 and Nasdaq 100. This concentration raises fundamental questions about how market structure, dependence, and systemic risk should be modeled in modern financial markets.

My work approached this problem from a network and graph-theoretic perspective. Rather than relying solely on traditional sector classifications or factor models, I represented equities as nodes in a correlation-based network and applied the Leiden community detection algorithm to identify cohesive groups of assets whose returns move together. This framework allowed me to quantify and visualize how market structure evolves over time, particularly across different market regimes.

Leiden Community Detection - Correlation Network Over Time

The video visualizes a dynamic stock correlation network over time using Leiden community detection. Each node represents a stock, while edges encode pairwise correlations, with greater width indicating stronger significance. Clusters of nodes correspond to detected communities. As time progresses, the animation illustrates how nodes and communities evolve and interact, highlighting regime changes in market structure. The M7 stocks are highlighted in gold.