基于改进遗传算法的证券投资组合研究

发布时间：2018-01-04 05:38

本文关键词：基于改进遗传算法的证券投资组合研究　出处：《东南大学》2017年硕士论文　论文类型：学位论文

更多相关文章： 遗传算法 CVaR 熵 多目标证券投资组合

【摘要】：随着我国经济的不断发展,我国的金融体系日益完善,投资者资产管理需求日益增加,推动各类金融产品迅猛创新发展。如何在保证收益的情况下降低风险,成为投资者越来越关注的焦点,但没有一种广泛有效的度量风险的方法和有效解决复杂优化问题的全局搜索算法。论文基于单目标CVaR模型和基于熵的多目标证券投资组合,应用改进遗传算法求解模型,探讨搜索能力更强且能跳出局部收敛的算法,选题具有一定的理论意义和较大的实用价值。论文首先综述资产组合理论、遗传算法的相关研究成果,分析熵理论运用到投资风险度量的合理性;接着针对实际投资行为对单目标CVaR基础模型进行扩展,通过改进遗传算法的交叉概率和变异概率在保持群体多样性的同时,保证遗传算法的收敛性;然后以东方财富漂亮50指数股票作为研究样本,使用改进遗传算法进行求解;最后以熵度量风险和CVaR作为目标函数,构成多目标证券投资组合,接着改进多目标遗传算法,使其能够预测未成熟收敛的产生。实证结果表明,论文构建的扩展模型和多目标优化模型有效的,且改进的多目标遗传算法能够优化模型的求解过程。
[Abstract]:With the continuous development of China's economy, China's financial system is increasingly perfect, investor asset management increasing demand, promote the rapid development of all kinds of financial innovation products. In order to reduce the risk income situation, has become the focus of investors pay more and more attention, but did not have a widely effective risk measurement method and the effective solution complex global optimization search algorithm. Based on the single objective CVaR model and multi-objective portfolio based on entropy, using improved genetic algorithm to solve the model, search ability is stronger and can jump out of local convergence of the algorithm, the topic has certain theoretical significance and practical value. The paper first summarizes the portfolio theory, correlation the research results of the genetic algorithm, analyses the rationality of entropy theory to the investment risk measurement; then according to the actual investment behavior based on single target CVaR The model was extended by improved genetic algorithm crossover probability and mutation probability in maintaining the diversity of groups at the same time, to ensure the convergence of genetic algorithm; and then to beautiful Oriental Fortune 50 stock index as the research sample, using the improved genetic algorithm; finally an entropy risk measure and CVaR as objective function, a multi-objective portfolio investment then, the improved multi-objective genetic algorithm, which can predict premature convergence. The empirical results show that the extended model established in this thesis and the multi-objective optimization model is effective, the process of solving the multi-objective genetic algorithm and improved optimization model.

【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F832.51

【参考文献】