A Topology-Driven CNN Framework for Stock Price Prediction Using News Sentiment and Market Dynamics

Stock price prediction remains a challenging task due to the volatile and non-linear behavior of financial markets, especially in emerging economies like Bangladesh. Existing models often fail to integrate both temporal patterns and structural features, limiting their predictive power. This study proposes a hybrid deep learning framework called TDACNN, which combines Topological Data Analysis (TDA) with Convolutional Neural Networks (CNN) to enhance prediction accuracy. The model utilizes rolling window inputs from stock price sequences, daily sentiment scores derived from financial news, and macroeconomic indicators including DSEX and DS30 indices from the Dhaka Stock Exchange (DSE). TDA features capture statistical trends across sliding windows, while CNN layers extract short-term local patterns. Optuna-based hyperparameter tuning is employed to optimize model configuration. The model is evaluated using data from five DSE-listed companies and compared with traditional and deep learning baselines such as BiLSTM, BiGRU, Random Forest, and Support Vector Machines. The experimental results show that TDACNN achieves better performance with lower MAE and MSE values and a higher R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score. This work demonstrates the advantage of combining topological and temporal features for improved stock prediction in emerging market environments.