This study uses daily data from February 1, 2020, to July 6, 2024, to analyse the dynamic connectedness of equities, bonds, energy, precious and base metals, cryptocurrencies, and agricultural and food commodities. We use the time-varying parameter vector autoregression (TVP-VAR) framework to assess aggregate and bilateral connectedness measures (TCI, NET, NPDC, PCI) and apply these findings to portfolio allocation using minimum variance (MVP), minimum correlation (MCP), and minimum connectedness (MCoP) techniques. The results show an average Total Connectedness Index of 42.54%, with sharp spikes during the COVID-19 crisis and the Russia–Ukraine war, underscoring the sensitivity of cross-asset spillovers to systemic shocks. Equities, notably the S&P 500 and QGREEN, perpetually act as net transmitters, whereas gold, BTC, and agricultural commodities predominantly behave as receivers. PCI analysis reveals stable clusters of strongly connected pairs – equities (S&P500–QGREEN), bonds (S500B–S500GB), and industrial commodities (copper–oil) – while agricultural assets (LC and SB) remain weakly connected in daily frequency, they become more integrated weekly, implying weaker diversification at medium horizons. Portfolio analysis demonstrates that MCP delivers the highest Sharpe ratio, with MCoP close behind, while MVP underperforms in periods of high equity–bond co-movement. Bilateral hedge ratios confirm that bonds are the most effective variance absorbers, but risk-adjusted outcomes improve when allocations also minimise connectedness.

Keywords