A Deep Approach for Forecasting the NSE Opening Index Employing CNN-LSTM Hybrid Framework

Forecasting share values has long been a formidable undertaking. Indeed, numerous analysts have a strong interest for study of share value forecasting. Predicting approaches classified as linear and non-linear methods. The prediction research offers a thorough assessment of the application of various soft computing approaches for analysing as well as predicting the opening index price value from the NSE. Here utilised networks are RNN, LSTM, CNN and CNN-LSTM hybrid model for forecasting future trends of opening price of NSE Index. The algorithms employed are trained on historical data that includes the initial and final costs, the maximum and minimum stock prices, and the volume of stocks purchased. The study specifically examines the daily closing costs of the NSE. The network’s algorithm is trained utilising both the historical price of the NSE index and the projected opening cost. Comparative analysis is conducted by employing various assessment criteria. The CNN-LSTM simulation establishes superior performance equated to other models. Through our investigation, we were able to determine the attributes choice influence as well as parameter optimisation for accuracy of predictions as well as attributes defined to assess the value of stock market. Prototypes’ performance was evaluated through Mean Squared Error statistic. This study examines that the statistical parameters RMSE (0.0051), MSE (0.0003), MAE (0.0032) is smallest and the value of R 2 (0.9997) is maximum for hybrid CNN – LSTM framework. The CNN - LSTM model exhibits lesser errors in comparison to the single LSTM, RNN and CNN framework.