D2CTNet: An Approach to Detect Dynamics of Candlestick charts and Technical indicators for Stock Market

Financial markets generate a great deal of data, including visual indicators, such as candlestick charts, and technical indicators extracted from historical data. In this work, we propose a deep learning model, the Detecting Dynamics of Candlestick charts and Technical indicators Network (D2CTNet), designed to capture and model the dynamics of visual and technical financial data. The visual component is derived from candlestick charts, providing insight into market trends and price movements. Moreover, technical indicators extracted from historical data provide quantitative information on the economic health of stocks. D2CTNet is an approach that learns the complex relationship between visual and technical indicators through a self-attention mechanism and improves the accuracy of stock price movement predictions. Validation on real data from stocks in the US, Chinese, Indian, and Korean markets demonstrates the efficacy of D2CTNet, outperforming existing models in terms of accuracy. This research presents a robust framework for integrating diverse financial features, paving the way for better insights into market behaviour and more accurate predictions. In addition, a user interface is developed to interact with the proposed D2CTNet model and provide trading actions based on the predictions made.