中文摘要：

从采购商角度出发，在同时存在现货、长期协议和期权多种采购渠道时，针对随机的现货价格和客户需求，建立了基于效用函数的采购商多阶段组合决策模型。通过数学推导给出了不同情况下最优采购决策所应满足的条件，使用遗传算法进行数值实验，并分析了市场环境对采购决策和收益的影响。在同一阶段，采购商的最优采购决策随着市场风险（现货价格或客户需求的波动程度）的增大表现为：长期协议采购量逐渐减少，通过逐渐增大预留给现货的采购缺口和期权预订量来降低数量风险；而在保持市场风险不变时，由于服从鞅过程的现货价格和客户需求会随着阶段递增逐渐放大风险，因此最优采购决策也表现为类似的变化趋势；相比单一现货采购方式，组合采购方式在不同市场环境都能获得更高和更稳定的收益。

英文摘要：

To deal with the uncertainty of stock price and customer demand, this paper establishes a multi-period portfolio decision model based on the quadratic utility function and from the perspective of a buyer, under the condition of stock market and contract market （including long-term and options）. The conditions that the optimal decisions satisfy in different circumstances are derived. Numerical experiments are conducted with Genetic Algorithm and the impact of market environment is analyzed. With the increasing market risk （spot price or customer demand variation）, the optimal decision of a certain period varies, the long-term contract volume diminishes gradually; the spot volume and the options reservation volume increase to mitigate quantity risk. However, the optimal decision has the same tendency over time with constant market risk, whose impact on the buyer would be augmented by the martingales of spot price and customer demand. The profit of portfolio strategy is higher and more stable than that of single spot strategy.