There are numerous stock market theories as a result of the gradual usage of mathematical models by researchers to forecast equities during the past few decades. By quantifying the rise and fall range, the prediction problem can be changed into a multi-classification problem based on the related data. This paper describes an Adversarial Learning-based stock forecast model by building a three-tier LSTM training network using the Adversarial Learning concept, selecting 300 stocks to represent the overall performance of the Chinese stock market, increasing the proportion of large to small training datasets, and strengthening the model’s ability to obtain detailed information from a small amount of data.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados