The Application of Artificial Intelligence to Stock Forecasting: A Literature Review

Due to the non-linearity, high volatility and noise characteristics of stock prices, the prediction of stocks has become a challenging issue. The results of stock prediction algorithms rely on the selected indicators, including financial indicators and market sentiment indicators, and the algorithm model. A large number of scholars have conducted studies and innovations from different perspectives respectively to optimize the prediction results. This paper reviews the development of artificial intelligence in stock application from two perspectives of index and algorithm model. Among them, the characteristics, advantages and disadvantages of 8 transformer models are shown, as well as the emergence of financial language models such as BloombergGPT and FinGPT. In addition, due to the particularity of China’s stock market, when making predictions about stocks in Chinese stock market, we are expected to focus on taking into account market sentiment, policy factors and adjusted financial indicators., so as to enhance the accuracy of prediction.