Threshold overnight comovement analysis of intraday and overnight returns

Abstract

This study develops a statistical framework to analyse lead-lag relationships between stock markets with non-overlapping trading hours. We examine the interdependence between markets with asynchronous sessions, such as those in the United States and Asia, by analysing intraday and overnight returns of representative exchange-traded funds. To quantify the direction, strength, and persistence of return comovements, we introduce a novel counting process, the Threshold Overnight Comovement process. A self-exciting intensity model is used to capture time-varying dynamics and momentum effects inherent in these relationships. Empirical findings reveal dynamic and reversal lead-lag patterns that become more pronounced during periods of elevated volatility. Trading strategies based on the proposed model demonstrate improved performance by capturing persistent and evolving cross-market return dependencies.