Author(s)

Junjie Guo

Corresponding Author

Junjie Guo

ABSTRACT

With the development of artificial intelligence, the core machine learning algorithms in artificial intelligence are becoming more and more mature in the application of finance. However, there are some common problems in most applications, such as over-reliance on native machine learning models, viewing the problem from a machine learning perspective rather than a financial perspective, and lack of optimization for financial applications. These problems inspire us to use and optimize models from a financial perspective. In this paper, we address the current problems in predicting stocks using machine learning for optimization and propose a formula to measure the degree of dispersion of prediction results called Dispersion Degree of the False Sample (DDFS). We give a quantitative and qualitative analysis of the optimization problem. The results indicate that our work can improve the efficiency of the model usage in quantitative trading and deepen the understanding of machine learning in finance.

KEYWORDS

Stock Prediction, Quantitative Trading, Machine Learning