Recurrent Neural Network-based Closing Index Prediction of Indian Software Industry Stocks

Stock market investors have continually searched for innovative and novel approaches that can help them accurately anticipate stock price movements and generate significant profits. Recurrent neural networks (RNN) of different types are utilized in time series analysis specifically for the prediction of stock prices. This work utilizes simple RNN, Long-Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) and Bidirectional RNN (BiRNN) to estimate the stocks of the top-five Indian Software industry stocks with different trends and various price ranges. We considered the TCS, Infosys, Wipro, HCL and Tech Mahindra stock datasets from 20/04/2018 to 15/09/2023. The closing indices of these companies are analysed with five-day time intervals, which are further scaled and fitted to RNN models. The performances of the deployed models are compared in terms of the mean squared error (MSE) and accuracy metrics. The results showed that the LSTM and BiRNN models produced lower MSEs and higher accuracies than did the other methods for the companies mentioned above since traditional RNNs have exploding or vanishing gradient problems.