Stock Index Prediction Using Temporal Convolutional Network and Long Short-Term Memory Network Optimized by Genetic Algorithm
This research proposes the use of GA (Genetic Algorithm), LSTM (Long Short-Term Memory) and TCN (Temporal Convolutional Network) to predict movements in the stock index with the aid of a hybrid model, and TCN efficiently extracts models that have long-term dependencies as well as local features from time-series data, whereas LSTM captures nonlinear relationships and long- term trends. GA optimizes the hyperparameters of the model to enhance the accuracy of prediction. The hybrid GA-TCN-LSTM model is applied to several international stock markets and shows superior forecasting performance and profitability compared to traditional neural network models. Our study aids in enhancing stock market forecasting with the employment of a robust model which can adapt to a broad variety of market conditions, improving investment decisions for financial analysts and investors.