RETRACTED ARTICLE: Research on pre- and post-fund performance prediction based on improved neural network algorithm

Abstract

At present, domestic and foreign scholars have systematic and perfect research on the sustainability of fund performance. However, the effect of fund forecasting still needs to be improved, and the conclusions on whether the fund performance is sustainable for different samples are quite different. This study constructs a fund prediction model based on improved neural network, uses genetic algorithm to adjust the initial weight and threshold of BP neural network, and comprehensively analyzes the parameters involved in neural network. Moreover, this study determines the optimal parameters through the training model and estimates the sustainability of the fund performance by introducing a neural network algorithm. The neural network algorithm can screen out systematic disturbances, analyze and calculate effective information, and can better screen out the effective parts of various fund performance persistence calculation methods, and does not need to set up a distribution function to take the significance test. In addition, this study selected the most appropriate network structure to take an empirical analysis.