AE-ACG : A Novel Deep Learning-Based Method for Stock Price Movement Prediction

We combine convolutional neural network (CNN) and gated recurrent unit (GRU) to form a new structure and embed it as a base layer in the autoencoder (AE) framework to efficiently extract features from financial time series data. To better utilize the hierarchical features, skip connection is employed to connect encoding and decoding process. In addition, attention mechanism (AM) is introduced to distinguish the importance of historical data in different periods. Extensive experiments show that the proposed model is effective in predicting price movements, showing advantages over traditional methods