Abstract
<jats:p>This study maps cross-asset information flow and network vulnerability across major equities, cryptocurrencies, commodities, and macro/FX indicators by translating Granger-causality relations into a directed network. Using daily data for 23 instruments (seven equity indices, five cryptocurrencies, six commodities, five macro/FX series) over 2020-04-13 to 2025-12-30 (1,485 observations), the sample is aligned to weekdays to avoid weekend-induced bias between 24/7 crypto and business-day markets before computing log returns. Stationarity is verified via a triangulation strategy (ADF, KPSS, and Zivot–Andrews). Pairwise linear Granger causality tests are then applied to all 506 ordered pairs (max lag 5), and statistically significant links are mapped to a directed graph. Network roles are quantified using in-/out-degree, betweenness, eigenvector centrality, and density, while time variation is assessed through rolling-window estimation (250-day window, 20-day step) and sub-period regime networks. The full-sample network contains 181 significant directed links (density = 0.358), indicating moderately high cross-asset connectivity. EURUSD and HSI are the most interconnected nodes (total degrees 25 and 24), SP500 is the strongest transmitter (out-degree 14), and HSI acts as the key bridge (betweenness 0.138). Rolling density ranges from 0.144 to 0.374 (mean 0.241; σ 0.055), implying regime-dependent spillover intensity and higher vulnerability when density exceeds 0.296. Sub-period density peaks in 2022 (0.283) and remains elevated in 2023–2025 (0.269). The results imply that diversification benefits weaken during dense regimes; monitoring high out-degree and bridge nodes can provide early warning of contagion pathways.</jats:p>