Prediction of Stock Trading Signal Based on Multi-indicator Channel Convolutional Neural Networks

Stock forecasting has always been a tempting and challenging problem in the field of financial research. Recently, convolutional neural networks have been used to classify trading signals of stock, but different indicator rankings will generate different images when generating data picture. A multi-indicator channel convolutional neural network (MICNN) is proposed to avoid the uncertainty of image generation. The test results are better than MLP and show that the proposed model has good classification performance, and the results of the simulated trading prove that the trading signals predicted by our method have practical value.