Analyzing Indian Stock Markets Through Correlations: Comparative Insights
Comparative analysis evaluating dependency measures across Indian equity sectors using empirical time-series correlation matrices.
Conference Presentation & Peer Review
This empirical research was formally presented during the technical sessions of the 3rd International Conference on Infrastructure Development and Sustainability (ICIDS 2025), hosted at Adani University, Ahmedabad. The work was published in the official conference proceedings by Atlantis Press / Springer Nature.
The podium presentation detailed the failure modes of standard linear correlation models during market turbulence—termed “The Linear Trap”—and defended the necessity of integrating multi-metric concordance measures for financial systemic risk evaluation. During the post-presentation Q&A session, we addressed peer inquiries regarding cross-sector volatility spillover, data cleaning for non-stationary market regimes, and the operational viability of lead–lag signals for institutional portfolio rebalancing.
Technical Methodology & Key Findings
Traditional risk architectures predominantly rely on Pearson linear correlation, assuming gaussian distributions and symmetric return dynamics. To overcome this limitation, our methodology establishes a multi-dimensional framework comparing four distinct dependency measures:
- •Pearson Correlation ($r$): Captures instantaneous linear co-movements across return vectors.
- •Spearman Rank Correlation ($\rho$): Evaluates monotonic dependencies without assuming normality or linear proportionality.
- •Kendall's Tau ($\tau$): Quantifies ordinal concordance, providing superior robustness to high-frequency market noise and heavy-tailed distribution anomalies.
- •Time-Lagged Cross-Correlation: Detects temporal phase shifts and directional transmission delays across sector pairs.
Sector-Wise Coupling & Auto Sector Matrix: A central empirical finding is the pronounced systemic synchronization within the Indian Automotive sector. As demonstrated in the correlation matrix below, automotive constituents exhibit high inter-stock coupling across all market cycles, forming a singular structural dependency block susceptible to joint systemic shocks.
In contrast to the tightly bound Auto sector, defensive sectors such as Pharmaceuticals displayed fragmented, low-magnitude correlation structures. This divergence confirms that linear-only risk calculations systematically overlook hidden non-linear coupling, underestimating portfolio drawdown risk by approximately 40% during adverse market regimes.