Crude benchmark oil prices play a crucial role in energy policy and investment management. Previous research confined itself to studying the static, uncertain, short-or long-term relationship between global benchmark oil prices, ignoring the time-varying, quantitative, dynamic nature of the relationship during various stages of oil price volatility. This paper proposes a novel approach combining grey relation analysis, optimization wavelet analysis, and Bayesian network modeling to explore the multi-period evolution of the dynamic relationship between global benchmark oil prices and regional oil spot price. We analyze the evolution of the most significant decision-making risk periods, as well as the combined strategy-making reference oil prices and the corresponding periods during various stages of volatility. Furthermore, we determine that the network evolution of the quantitative lead/lag relationship between different influences of global benchmark oil prices shows a multi-period evolution phenomenon. For policy makers and market investors, our combined model can provide decision-making periods with the lowest expected risk and decision-making target reference oil prices and corresponding weights for strategy adjustment and market arbitrage. This study provides further information regarding period weights of target reference oil prices, facilitating efforts to perform multi-agent energy policy and intertemporal market arbitrage. (C) 2016 Elsevier Ltd. All rights reserved.