Analysis of Stock Market Quantitative Trading Strategies Based on Machine Learning

From the issue of the first stock so far, China's stock market has gone through 35 years of development. In the 21st century, influenced by modern technological concepts such as artificial intelligence, cloud computing and big data, the stock market has put forward quantitative investment methods with machine learning as the core, which have been widely used in the field of stock investment. The United States and other developed countries use quantitative tools to improve the securities market, not only achieve more economic benefits, but also get rid of the limitation of traditional investment mode in practice. Although at present our country securities market increasingly perfect, data analysis and processing power is more and more strong, more abundant with quantitative investment research data resources, the basis of practical investment development condition is increasingly mature, but from the overall perspective, quantify the level of investment in our country is still at the primary stage, the actual investment tools can only produce auxiliary role in the activity. This paper takes 300 components as an example, selects the trading day factor data from January 2015 to December 2019, compares the three common machine learning classification algorithms: random forest, extreme gradient promotion and support vector machine, and discusses the potential relationship between the factor data and real yield of stocks of listed companies. The final results show that the classification algorithm of support vector machine has higher performance than the other two algorithms in the process of quantitative research.