A Multi-Strategy and Business Analysis Strategy-Enhanced Most Valuable Player Algorithm for Global Optimization and Corporate Bankruptcy Prediction

The Most Valuable Player Algorithm (MVPA) is a recently developed metaheuristic optimizer with a simple competition-based framework; however, its search capability is limited by insufficient information interaction and weak diversity maintenance. To address these issues, this study proposes a Multi-Strategy Enhanced Most Valuable Player Algorithm (MSEMVPA). Three complementary strategies are developed: an Adaptive Historical Differential Competition Strategy (AHDCS) that introduces historical search information and adaptive differential guidance to enhance exploration, an Adaptive Multi-Elite Reorganization Strategy (AMERS) that integrates diverse elite information to improve exploitation, and a Business Analysis Strategy (BAS) that reconstructs inferior individuals to maintain population diversity. The proposed MSEMVPA is evaluated on the CEC2017 benchmark suite through comparative experiments, ablation studies, convergence analysis, statistical tests, and computational complexity analysis. Furthermore, MSEMVPA is employed to optimize a multilayer perceptron model for enterprise bankruptcy prediction. Experimental results demonstrate that MSEMVPA achieves improved optimization accuracy and robustness compared with the original MVPA and several competitive algorithms. The results indicate that the proposed multi-strategy framework provides an effective approach for enhancing MVPA and solving complex optimization problems.