Competitive feedback particle swarm optimization enabled deep recurrent neural network with technical indicators for forecasting stock trends

The stock market prices are dynamic, thus remaining a major challenge in forecasting future stock trends. The Competitive Feedback Particle Swarm Optimization-based Deep Recurrent Neural Network (CFPSO-based Deep RNN) is created to ensure an efficient forecast of the stock market. The forecasting is done concerning the precedent and up to date status of the market. First, the input is submitted to the features extraction phase to extract technical indicators, and then the extracted practical indicators are used to forecast stock market movements. In addition, feature fusion and the data augmentation process effectively enhance the prediction quality. Finally, the Deep RNN classifier is accomplished in the forecast module, where the preparation method of the Deep RNN is performed using a developed optimization algorithm named CFPSO. The developed CFPSO is planned by hybridizing the Competitive Swarm Feedback Algorithm (CSFA) and Particle Swarm Optimization (PSO). The implementation of the proposed work is done in PYTHON. The developed CFPSO-based Deep RNN exhibits superior performance based on MAE, MSE, RMSE, accuracy, sensitivity and specificity with values of 0.136, 0.107, 0.246, 0.963, 0.957 and 0.980, respectively.