Optimization and BP Neural Network Prediction with Correction Based on Quantitative Stock Trading Models

In this paper, a new quantitative stock trading model is proposed and an optimization problem is formulated for investment returns. By solving this optimization problem using the Neider-Mead algorithm, the optimal returns and their corresponding decision variables, that is, the optimal trading coefficients, are obtained for a fixed time period. However, since these trading coefficients are derived based on historical data, they may not work well in future investments. Therefore, this paper uses BP neural network to predict the current trading coefficients and corrects the predicted trading coefficients by a correction model in order to apply them to future stock invest-ments. Finally, the corrected trading coefficients are substituted into the quantitative stock trading model, and the real stock arithmetic simulation is performed according to the established quantitative trading rules.