A Quantitative Investment Model Based on Random Forest and Sentiment Analysis

Abstract

In recent years, under the influence of economic globalization and anti-globalization, the stock market has experienced great fluctuations in China. Quantitative investment has attracted a lot of attention because of its characteristics of maintaining stable returns. Existing research is unilaterally based on quantitative data or qualitative data for analysis to construct a quantitative investment model. This paper considers both quantitative and qualitative data to construct a more comprehensive model than that in the past. Based on the optimized database, we present a combinational model named RF-SA, which is composed of random forest and sentiment analysis model. First of all, this paper uses the SBS algorithm to select the characteristics of stock transaction historical data, optimizes the prediction database, reduces data redundancy, and improves the accuracy of the model. Secondly, we analyze the characteristics of the Chinese stock market and study the advantages and disadvantages of many data mining algorithms, and select random forest model, the most suitable model, to build the first step of stock selection model. Then, through the analysis of public opinion, the confidence index of the stockholders is calculated; on this basis, the results of the RF model and the confidence index are combined to make a second choice for the stock, and the quantitative investment portfolio is obtained, and excess returns can be obtained. The results of empirical data show that, the RF-SA model obtains a higher rate of return than the investment model of the Shanghai Stock Index.