Information diffusion through weighted positive causal networks: Evidence from pair-based trading strategy in China

This study develops a novel weighted positive causal network (WPCN) approach to analyze directional information flow and causal return comovement in stock markets. Building on Granger causality tests, the proposed method distinguishes positive lead-lag relationships—reflecting differences in information diffusion speed—from negative ones that capture heterogeneous responses to shared information. Using daily data for the CSI 1000 constituents from 2019 to 2023, we construct positive causal networks and evaluate their predictive power through a pair-based trading strategy. Empirical results show that the positive causal network effectively identifies stable positive lead-lag relationships among small- and mid-cap stocks in China. Trading strategies based on the WPCN achieve higher winning rates and cumulative returns than those relying on unweighted causal networks, particularly for strongly positive causal linkage stock pairs. These findings highlight the economic significance of causal linkages in return comovement and provide new evidence on the dynamics of information transmission in China's small- and mid-cap stock market.