Meta-algorithm And Its Application To Portfolio Selection

With the continuous development of science and technology,the interpenetration of technologies in different fields is conducive to solving the problems existing in each field.Among them,the development of computer technology has made a great contribution to promoting the development of China's financial market and solving the academic problems in the financial field.In recent years,"smart" marks a noun,such as machine learning techniques are active and widely used in the field of portfolio management,among them,the integrated study of machine learning techniques for portfolio selection problem solution and technical support,integrated learning could be combined with different base algorithm,the method by combining multiple learning effect of poor weak learning good strong learning learning effect is obtained.Meta-algorithms in this paper make choice of portfolio selection problem converted to classification problem,training investment model to predict which stocks will obtain higher yields in the next invest period,investors can choose the more believable stocks according to the results of the model recommended,so as to solve the question that the choice of stocks before the investment decision.This paper applied two meta-algorithm,which are the Ada Boost algorithm and the Random Forest algorithm.On the one hand,it improve the process and the advantages of the meta-algorithms from the angle of the theory.On the other hand,this article proves that the models which are obtained by the two meta-algorithms in the performance of the test sample set in cumulative yield,average annual return and variance are satisfied,which proves that the application of the two meta-algorithm to solve the problem of portfolio choice has practicability and superiority from the perspective of numerical analysis.