Abstract: With the improvement of people’s living standards, increasing numbers of people are involved in the stock market. To scientifically and effectively quantify stock selection, this study establishes a quantitative stock selection model through an organic combination of quantitative investment, deep learning, and text analysis. That is, stocks with good fundamentals are selected through text analysis. Then, those with good prediction accuracy are selected using long-short term memory (LSTM). Finally, the stock price trend of these stocks in the next few days is predicted. In terms of empirical analysis, the model is used to perform calculations on some stocks and the result shows Yinghe Technology with better forecasting effects.

Key words: quantitative stock selection, text analysis, long-short term memory (LSTM), forecast