改进遗传算法在投资组合中的应用.doc

摘 要 遗传算法起源于对生物系统所进行的计算机模拟。美国密执安大学的Holland教授及其学生受到这种生物模拟技术的启发，创造出了一种基于生物遗传和进化机制的适合于复杂系统优化计算的自适应概率优化技术---遗传算法。证券投资组合优化问题的实质就是有限的资产在具有不同风险收益特性的证券之间的优化配置问题。 因此，本文根据上述要求把交易成本和股票的整手买卖引入含有风险偏好的Markowitz组合投资模型，并对证券组合进行分类约束来降低风险，从而构造了含有约束的混合整数非线性规划模型。 遗传算法是一类模拟自然界生物进化过程与机制，求解问题的自组织和自适应的人工智能技术。由于其运行简单和解决问题的有效能力而被广泛应用到众多领域。但是它也容易产生早熟现象以及局部搜索能力比较差，所以对很多问题而言，基本遗传算法并不是解决问题的最有效方法。因此本文对基本遗传算法的一些算子进行了改进，获得了较满意的结果。 本文提出的组合投资模型在求解上存在一定的难度，采用遗传算法求解。在计算机上用Matlab7.0编程实现。 关键字：遗传算法；生物模拟；投资组合；交易成本； ABSTRACT Genetic algorithm originated in biological systems through the computer simulations. Holland Michigan University professor and his students are subject to this biological simulation technology inspired to create a bio-based genetic and evolutionary optimization of complex systems for adaptive probability calculation --- genetic algorithm optimization technique. Portfolio Optimization essence of the problem is the limited assets with different risk and return characteristics of the optimal allocation between the securities issue. Therefore, this paper according to the requirements of the transaction costs and stocks containing whole lot introducing risk appetite Markowitz portfolio model, and classify constraints portfolio to reduce risk, which is constructed with constrained mixed-integer nonlinear programming model Genetic algorithms are a class of simulation of natural biological evolution and mechanisms for solving the problem of self-organization and adaptive artificial intelligence technology. Because of its operational simplicity and ability to solve problems effectively been widely applied to many fields. But it is also prone to premature and relatively poor local search ability, so many problems, the basic genetic algorithm is not the most effective way to solve the problem. This article on some of the basic genetic algorithm has been improved operator to obtain a more satisfactory result. The proposed model for portfolio investment in the solution th