Stock Recommendation Based on Depth BRNN and Bi-LSTM

In the face of increasingly complex stock data information, the prediction model based on traditional statistical model is gradually unable to meet the needs of user personalization, accuracy and short time. A prediction model based on deep bidirectional cyclic neural network (BRNN) and bidirectional long and short term memory network (Bi-LSTM) was proposed to make the prediction data more reasonable for users to make decisions. This model introduces deep neural network and Dropout algorithm, which has strong anti-jamming ability and learning ability. It not only solves the problem of gradient disappearance and gradient explosion caused by long-term dependence of learning, but also optimizes the problem of overfitting and slow convergence caused by deep network model to some extent. The experiment on the NSE data set shows that this prediction model improves the determination coefficient by 5% and reduces the error by 3%-5% compared with the existing prediction model.