Mutual fund grouping has been widely applied in financial services. These applications are based on the premise that mutual funds within the same group exhibit similar performance, and this similarity will persist over subsequent periods. However, in practical applications, significant variations in the performance of mutual funds within the same group have been observed, negatively impacting higher-level applications based on mutual fund grouping. To address this challenge, we propose an integrated framework comprising a highly intuitive grouping method which clusters mutual funds with a keen focus on their performance pattern, and a theoretically robust assessment of group persistence. To validate the proposed framework, we construct an experimental dataset comprising open-end equity mutual funds that were managed by single fund managers from January 2022 to June 2024 to ensure consistency in investment style. Experimental results demonstrate that our grouping method outperforms the widely adopted nine-grid Style Box approach and representative performance-based grouping methods. Furthermore, we observed a significant correlation between group persistence and several factors, including the overall market environment, the granularity of grouping, and the duration of period. To the best of our knowledge, this is the first study that quantitatively examines persistence of mutual fund grouping which bears fundamental import for practical applications. The code and dataset for this project are publicly available ( https://github.com/MengZou0905/FGPA ).