An Optimal Medium- to Long-Term Investment Strategy Using Kernel Method and Control Policy

本研究利用非线性控制策略来动态地重新平衡投资组合，以进行中长期资产配置。利用非线性数据分析的一类算法--核方法，可以将问题归结为一个凸二次优化问题。此外，基于特征值分解的降维技术也被开发，以减少问题的规模。最后以数值实验来评估本研究所建立策略的投资绩效，以及降维技术的有效性。

The present study utilized nonlinear control policies to dynamically rebalance the portfolio for medium- to long-term asset allocation. By using the kernel method which is a class of algorithm for nonlinear data analysis, the problem can be reduced to a convex quadratic optimization problem. Moreover, a dimensionality reduction technique based on eigenvalue decomposition was also developed to reduce the problem size. Numerical experiments were conducted to assess the investment performance of the strategy established by this study and the effectiveness of the dimensionality reduction technique.