With the expansion of the global economy, the stock market has rapidly developed. Due to its easy accessibility, diverse possibilities, and the potential for lucrative returns, stock trading has become the most attractive financial instrument for investors. However, investing in stocks typically involves high risks, as stock prices are influenced by various reasons, including trader sentiments, financial news, global economic conditions, national industry trend, and geopolitical events. Therefore, the ability to accurately predict stock prices is crucial for investors. Fuzzy theory can effectively address the data uncertainty issues in financial markets. Multi-task learning enables the simultaneous learning of multiple relevant tasks, accepting those tasks to share knowledge during the learning procedure. It applies the correlation within multiple tasks to increase the model’s generalization ability across tasks. Parametric margin-based support vector machine is suitable for handling data with heterogeneous structural noise. This paper integrates the advantages of these methods, and proposes a novel multi-tasks multi-kernel fuzzy support vector machine with parametric margin model for the prediction of movement of stock price.