Comparison of garch model and artificial neural network for mutual fund's growth prediction

摘要

The trend of investment has moved towards open ended funds, which removes the burden of investment from investors and promise certain percentage of profit. An open-end fund is a specialized type of mutual fund through which an investor can invest at any time. This kind of funds buy and sell shares as per their Net Asset Value per unit (NAV per unit). The freedom of time for investment is a big plus for such funds. There is more vigilance/security required for open-end funds. Research tries to build prediction models based on publically available data of Asset Management Companies (AMCs) and predict the growth of funds based on the time series analysis. The data includes past ten years data of top 9 AMCs, which is pre-processed to build a model for prediction of price/value for both individual investors and AMCs. Generalized Auto-Regressive Conditional Heteroskedasticity (GARCH) and Artificial Neural Network (ANN) are applied separately on the data to predict the NAV for next five months. GARCH model gave predictions with a very less Mean Square Error (MSE), outperforming ANN with a significant difference.