Research on Forecasting Volatility Rate of Financial Market based on GARCH and Neural Networks

Abstract

Aiming at the problem of irregular volatility in financial markets, this paper studies prediction algorithms based on GARCH class model and BP neural network model. In this paper, we compare the prediction accuracy between three kinds of GARCH base models and seven neural networks models. Results show that GARCH class model has a new neural network which can enhance volatility prediction accuracy; GARCH class model better, its variables as input value to neural network combination model help larger; additional introduction similar model variables reduce neural network combination model performance unless the new model variable brings information improvement advantages cover its data redundancy shortcomings.