Exploring Deep Learning for Stock Price Forecasting: A LSTM-RVFL-Attention Framework

This work proposes a hybrid LSTM-RVFL-Attention model for forecasting stock prices by combining LSTM, RVFL, and Attention. It learns temporal dependencies, features learning optimization, and focuses attention on significant time steps to maximize accuracy. Tried on six data sets, including stock data, index data, and futures data, it outperforms standard models with minimum error rates and maximum R² score. However, its lack of external influences suggests avenues for future improvement.