Sports performance improvement and prediction contains the analysis of various factors influencing athletic performance, including player statistics, team dynamics, injuries, and environmental conditions. Challenges of traditional methods in sports performance improvement and prediction include data privacy concerns, over fitting issues, complexity, and interpretability. To overcome these complexities, this paper proposed a novel method named the Adaptive Convolutional Encoder-decoder-based Gooseneck Bernacle Search (ACED-GBS) algorithm. In this study, a Convolutional Neural Network (CNN) is utilized to extract data related to athletes’ sports performance. Additionally, the encoder-decoder is employed to efficiently capture the interactions between the information. In this work, the Gooseneck Bernacle optimization with initial search strategy is implemented for hyperparameter optimization to enhance the performance of the ACED-GBS method and the study conducted experiments on the ODI-Players performance dataset. Different evaluation metrics namely precision, accuracy, recall, F1-score, specificity, etc are utilized to evaluate the performance of the ACED-GBS method and compare its performance with existing methods. The experimental outcomes depict the effectiveness of the ACED-GBS method for sports performance improvement and prediction. The experimental analysis demonstrates that the convergence speed of the ACED-GBS method is high, the error is low and the prediction performance is more accurate with high noise immunity and practicality compared to traditional methods.