Deep Learning-Driven International Market Trend Prediction and Trade Strategy Optimization

Abstract The rapid development of global trade and economic integration has heightened the need for accurate international market trend prediction to inform trade strategies. Traditional forecasting methods often struggle to capture the complex temporal and spatial relationships in global trade data. To address this challenge, this study proposes a deep learning-driven international market trend prediction model based on the ATT-CNN-LSTM framework, integrating convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and an attention mechanism. The CNN component extracts spatial dependencies among multidimensional trade indicators, while the LSTM component models the temporal evolution of trade patterns. The attention mechanism further enhances the model by assigning greater weights to influential time steps, reducing noise from irrelevant data points. Additionally, the AdamW optimization algorithm is employed to enhance training efficiency and model generalization. Experimental validation on an extensive dataset of global trade transactions demonstrates that the proposed ATT-CNN-LSTM model significantly outperforms conventional prediction techniques. The results indicate superior predictive accuracy and robustness in capturing intricate market dynamics. The findings of this study provide valuable insights for policymakers and business strategists in optimizing trade decision-making and mitigating market uncertainties. This work highlights the potential of deep learning in enhancing the precision of international trade forecasting and developing data-driven trade strategies for global economic sustainability.