在资产收益可预测的金融市场中,研究了连续时间最优动态资产配置问题. 利用分离定理,带有预测变量的优化问题被分解为一个参数推断问题和一个随机优化问题,
利用拉格朗日对偶方法和动态规划方法求得了最优策略和有效前沿.结果表明,预测变量带来的估计误差和投资机会集合时变性都会对最优策略和有效前沿产生显著影响.

This paper considers a dynamic asset allocation problem when the asset return is predictable in a continuous-time financial market. By using the separation principle, the dynamic portfolio optimization problem with predictable asset return is decomposed into two separate problems: a parameter inference problem and an optimization problem. The analytic expressions of the optimal portfolio strategy and the efficient frontier are derived by using the Lagrange duality approach and the dynamic programming approach. The result shows that both the estimate errors of the predictable variable and the time-varying investment opportunity caused by the predicative variable will lead to substantial impacts on the optimal strategy and the efficient frontier.