大数据背景下遗传算法在投资组合优化中的效果研究

Directly Evaluating Genetic Algorithms for Portfolio Optimization in the Context of Big Data

十八大报告以及十八大三中全会再三提到了发展多层次资本市场,进一步分散金融风险,促使我国对投资组合的应用更加广泛,更是对投资组合模型的有效边界的研究提出了新的要求。近些年来,随着计算机科技的不断发展,遗传算法在不同领域得以广泛应用,在投资组合领域也得到了广泛重视。为了科学系统的探索遗传算法在投资组合模型中的应用效果,本文设计了从5,10,50,100,200,300,400只股票,结合标准投资组合模型、带上界约束条件的投资组合模型以及带市值约束条件的投资组合模型三种约束条件。应用遗传算法对不同投资组合的有效边界分别进行求解,进而与由参数二次性规划法求得的有效边界的精确解进行比较,研究不同数目、不同约束条件下遗传算法的有效率。分析结果表明,在三种约束下,随着股票数目的上升,遗传算法的有效率呈逐渐下降的趋势。多目标投资组合成为了一个活跃的领域,我们也提出了相应的对遗传算法效果的评估与建议。

It has been discussed for several times to develop the multi-level capital markets,further diversify the financial risks,broaden the application of portfolio and raise new requirements for the research on efficient frontier in the Eighteenth National Congress of the Communist Party of China and the Third Plenary Session of the 18 th CPC Central Committee.In recent years,as the development of computer technology,Genetic Algorithms are widely used in different fields and also attached to more importance in the realm of portfolio theory.To explore the effectiveness of GA in portfolio optimization,this paper designs5,10,50,100,200,300,400 stocks are respectioely dengned,combining three constraint types of standard portfolio model,portfolio model with upper bound constraint and portfolio model with market value constraint.In this study GA is used to solve the efficient frontier of different portfolio and then compare with the exact solutions from parametric quadratic programming method,and the effectiveness of GA is evaluated with different numbers and different constraints.The results show that,in three constraints,as the number of stocks rising,effectiveness of GA shows a decreasing trend.Multi-objective portfolio has become an active field,and the assessment and suggestions are also put forward for the effects of GA.