Forecasting Stock Price Index Volatility with LSTM Deep Neural Network

Abstract

In strong noisy financial market, accurate volatility forecasting is the core task in risk management. In this paper, we apply GARCH model and a LSTM model to predict the stock index volatility. Instead of historical volatility, we select extreme value volatility of Shanghai Compos stock price index to conduct empirical study. By comparing the values of four types of loss functions, we illustrate that LSTM model has a better predicting effect.