<p>This research introduces a novel framework for identifying and exploiting predictive lead-lag relationships between equities. It proposes an integrated methodology that synergises advanced statistical techniques with machine learning models to enhance the detection and capitalisation of these predictive linkages. A Gaussian Mixture Model was employed to cluster nine major stocks according to their mid-range historical volatility profiles across a three-year dataset. Based on these clusters, a multi-stage causal inference pipeline was constructed, which incorporated the Granger Causality Test, a customised Peter-Clark Momentary Conditional Independence test, and Effective Transfer Entropy to isolate robust predictive connections. Subsequently, Dynamic Time Warping and a K-Nearest Neighbours classifier were utilised to ascertain the optimal time lag for trade execution before the strategy was rigorously backtested. The developed volatility-based trading strategy, evaluated from 8 June 2023 to 12 August 2023, demonstrated considerable efficacy. The portfolio achieved a total return of 15.38%, substantially outperforming the 10.39% return of a comparative Buy-and-Hold strategy. The strategy’s viability was further confirmed by key performance indicators, including a Sharpe Ratio of up to 2.17 and win rates reaching 100% for specific asset pairs. This work contributes a systematic and robust methodology for uncovering profitable trading opportunities from volatility-based causal relationships. The findings hold significant implications for both academic research in financial modelling and the practical implementation of algorithmic trading, presenting a structured approach for the development of resilient, data-driven strategies.</p>