A neural network approach to mutual fund net asset value forecasting

In this paper, an artificial neural network method is applied to forecast the end-of-year net asset value (NAV) of mutual funds. The back-propagation neural network is identified and explained. Historical economic information is used for the prediction of NAV data. The results of the forecasting are compared to those of traditional econometric techniques (i.e. linear and nonlinear regression analysis), and it is shown that neural networks significantly outperform regression models in situations with limited data availability.