•The study proposes a new network topology approach based on the STVAR model.•We use the wavelet packet decomposition to analyze multi-scale risk spillovers.•The systemic risk spillovers among commodities are asymmetric under different conditions.•The center of risk spillover networks varies across the market conditions and frequencies. This paper proposes a new network topology approach based on the STVAR model to identify asymmetric impacts of market conditions on multi-scale systemic risk spillovers of commodity markets. The results show that bearish market conditions enlarge low-frequency systemic risk spillovers in commodity markets, and bullish market conditions have more striking impacts on high-frequency systemic risk spillovers. Furthermore, the center of risk spillover networks varies across the market conditions and frequencies. Specifically, at the high-frequency level, sugar is the largest risk transmitter in bad regimes, and heating oil is in the center of the network in good regimes. At the intermediate frequency level, soybean becomes a more important risk transmitter in both regimes. In other cases, heating oil is the center of risk spillover networks.