LSTM-Integrated Particle Swarm Optimization Model for Listed Companies

Abstract

In the context of global economic integration, credit risk assessment of listed companies is crucial for maintaining capital market stability. However, traditional methods rely on static financial indicators, which are limited by timeliness and subjective parameter selection. While LSTMs excel at capturing long-term dependencies in financial time series data, they are limited by the difficulty in accurately determining key hyperparameters, such as the number of hidden layer neurons and the learning rate, which hinders their predictive performance. Therefore, this study proposes a credit risk assessment model for listed companies based on a combination of a long short-term memory network (LSTM) and a particle swarm optimization (PSO) algorithm. This model overcomes the core limitations of traditional methods in time series data processing and parameter optimization. Using deep learning techniques, the model automatically extracts features from time series data and effectively captures the dynamic impact of different financial indicators on credit risk. The particle swarm optimization algorithm is used to optimize the LSTM model's hyperparameters, thereby improving predictive accuracy and stability. Empirical results show that the proposed model outperforms traditional methods across multiple evaluation metrics, achieving an accuracy of 92.3% and a recall of 90.1%, validating its effectiveness and practical value in credit risk assessment. This model theoretically overcomes the bottleneck of LSTM parameter tuning. In practice, it can provide investors and regulators with a high-precision risk assessment tool, helping to improve capital market risk management.