Research on Baijiu Stocks: Integrating Entropy-Weighted TOPSIS and LSTM

Abstract

This study addresses the significant stock price volatility in China’s Baijiu industry and the inefficacy of traditional prediction models by proposing an intelligent forecasting framework integrating entropy-weighted TOPSIS and LSTM neural networks. A 17-dimensional evaluation system is constructed, encompassing price, trading, valuation, technical indicators, and financial data. The entropy-weighted TOPSIS method identifies Kweichow Moutai (600519.SH) and Shanxi Xinghuacun Fenjiu (600809.SH) as target stocks. Subsequently, an LSTM model is employed for multi-timescale price prediction, demonstrating superior accuracy compared to conventional models, particularly in short-term forecasting. Empirical results validate the framework’s effectiveness in capturing nonlinear trends and mitigating overfitting. The research offers actionable insights for investors and regulatory strategies, advancing intelligent decision-making in volatile financial markets.