This research examines the evolution of market microstructure at the National Stock Exchange of India (NSE) from 2020 to 2024, a period characterized by substantial growth in algorithmic trading from 35% to 44% of total trading volume. Using market microstructure data and analytical techniques grounded in complex systems perspectives, the study documents temporal patterns in price discovery, liquidity, volatility, and market efficiency associated with this digital transformation. The analysis reveals several notable changes in market characteristics. Transaction costs improved significantly, with bid-ask spreads declining by 23.4% and market depth increasing by 18.1%. Price adjustment half-life decreased by 50%, indicating faster information incorporation. However, market efficiency dynamics exhibited non-linear progression. Hurst exponent analysis shows markets initially moved toward random walk behavior (0.524 to 0.503, 2020-2022) before shifting to mean-reverting patterns (0.495-0.487, 2023-2024), challenging assumptions of monotonic efficiency gains. Network clustering coefficients increased by 23.5%, potentially reflecting strategy homogeneity rather than systemic interconnection. GARCH modelling revealed increasing volatility persistence, with parameters approaching unity. While average liquidity improved, liquidity variability during stress periods increased by 34%, and flash-event-like occurrences were more frequent during high algorithmic activity periods. Critically, these patterns represent temporal associations rather than established causal relationships. Multiple confounding factors—including the COVID-19 pandemic, macroeconomic volatility, and regulatory changes—prevent definitive causal attribution. The findings underscore the complexity of digital market transformation in emerging economies and highlight the need for enhanced monitoring, adaptive regulatory frameworks, and further research employing stronger identification strategies to understand the mechanisms underlying observed market evolution.
Ahirrao et al. (2026) studied this question.