Technical Indicator-Driven Deep Neural Networks for Stock Market Forecasting

Predicting stock prices is a tough task that is affected by multiple factors in the market and past trends. This research analysis investigates deep learning techniques for predicting India’s Nifty 50 stock price using historical data. The dataset includes the daily stock price values along with important technical indicators, which help increase the precision of the prediction. Multiple deep learning architectures, such as LSTM, CNN, LSTM, CNN-BiLSTM, CNN-LSTM with Attention, and Transformer-LSTM, are being assessed. With an RMSE of 0.028, the CNN_BiLSTM model had the lowest error, demonstrating a 26 % to 36 % decline in RMSE compared to the other models. The results demonstrate the efficacy of sophisticated deep learning architectures along with technical indicators in forecasting stock performance. It shows that hybrid models perform better than conventional networks.