Achieving superior stock price predictions with hybrid LSTM models and optimization techniques

Stock price forecasting remains a significant challenge in volatile financial markets. This study presents a novel hybrid model, the Long Short-Term Memory Variational Mode Decomposition Lemur Optimizer, which improves prediction accuracy and robustness. This model effectively balances exploration and exploitation during optimization by incorporating Lemur Optimizer's unique leaping and dancing behaviors into Long Short-Term Memory networks. Addressing limitations in prior methodologies, such as Long Short-Term Memory Sin-Cosine Algorithm Autoregressive Integrated Moving Average Generalized Auto Regressive Conditional Heteroskedasticity and Long Short-Term Memory Variational Mode Decomposition Rabbit Optimization Algorithm, the proposed approach captures both linear and nonlinear patterns in financial data. The model performed better than existing models when tested against historical stock data across ۱۳ prominent datasets and across different financial instruments. The research provides a robust framework for enhanced stock price predictions that facilitates more informed investment strategies and improved risk management in the field of financial forecasting.

کلیدواژه ها:

Financial Forecasting ، Lemur optimizer ، Variational Mode Decomposition ، Forecasting Precision