Research on ESG Excess Returns Based on Machine Learning and Cloud Computing

Value stocks have higher returns than growth stocks. This project intends to use Stacking technology to construct DLSB-Stacking models through the integration of multiple machine learning models. The purpose of this project is to use computer classification algorithm to construct various types of investment portfolios in China a stock market, and to conduct empirical analysis on the respective return characteristics of Shanghai and Shenzhen stock markets. The results show that the return rate of value stocks is higher than that of growth stocks in Chinese stock market, and this effect can only be reflected in small stocks in Shanghai stock market. But the outsized returns of small-cap stocks are concentrated in value stocks in Shanghai and value and hybrid stocks in Shenzhen. This paper also applies the DLSB-Stacking and Fama-French three-factor models to the empirical research of the Chinese stock market. In quantitative trading, the classical machine learning method represented by BP neural network can accurately grasp the actual situation in the market, so as to build an efficient securities portfolio. The description of stocks in a single stock model is very limited, but the integration of a Stacking model can better capture more market information than a separate model, thereby achieving higher excess returns.