Applying Random Forest Algorithm and Mean-Variance Model in Portfolio Optimization in the China Stock Market

Abstract

Machine learning has been used widely in robots and artificial intelligence (AI). Machine learning and deep learning significantly have contributed to the development of AI in recent years. AI was proposed in 1950 but due to the limitations of hardware and low computational efficiency, it gradually was introduced in the 1980s. However, the advancement of hardware, the considerable improvement in computer processing speed, the learning ability, and the efficiency of machine learning have led to fast growth surpassing human ability in many fields. Machine learning can find the logical relationships between explanatory and explained variables in large amounts of data. Therefore, the financial market is suitable for the application of machine learning. Current applications of machine learning in the financial industry include predicting bond defaults, suspicious transaction forecasting, analyzing suitable loan amounts, predicting and recommending suitable investment suggestions for customers, and investment. Based on this, using historical data such as the trading volume, return, turnover, price-book ratio, price-to-sales ratio, and a price earning ratio of the constituent stocks of the CSI 300 Index, and employing the random forest algorithm in machine learning, we established an investment portfolio strategy for stocks to predict the future price increase and probability of an increase. The mean-variance model was used, and the results showed that the performance of these two types of investment strategies significantly outperformed the CSI 300 Index.