A Stock Price Prediction Approach Based on Time Series Decomposition and Multi-Scale CNN using OHLCT Images
This addresses stock prediction for investors in the China A-share market, but it is incremental as it builds on existing image-based deep learning methods.
The paper tackled stock price prediction in the China A-share market by proposing a novel CNN-based method that combines sequential and image features, achieving a 61.15% positive predictive value, 63.37% negative predictive value, and 165.09% total profit over 4,454 stocks.
Recently, deep learning in stock prediction has become an important branch. Image-based methods show potential by capturing complex visual patterns and spatial correlations, offering advantages in interpretability over time series models. However, image-based approaches are more prone to overfitting, hindering robust predictive performance. To improve accuracy, this paper proposes a novel method, named Sequence-based Multi-scale Fusion Regression Convolutional Neural Network (SMSFR-CNN), for predicting stock price movements in the China A-share market. By utilizing CNN to learn sequential features and combining them with image features, we improve the accuracy of stock trend prediction on the A-share market stock dataset. This approach reduces the search space for image features, stabilizes, and accelerates the training process. Extensive comparative experiments on 4,454 A-share stocks show that the model achieves a 61.15% positive predictive value and a 63.37% negative predictive value for the next 5 days, resulting in a total profit of 165.09%.