Modeling and predicting stock market volatility using neural network and conditional variance patterns

Modeling and predicting stock market volatility using neural network and conditional variance patterns The fluctuation forecast is one of the most important issues in the financial markets, which attracted the attention of many academic researchers and experts in the field over the past few decades. In this study, considering this necessity, we examine the modeling and prediction of stock market volatility using the combination of artificial neural networks and conditional variance patterns. In this research, multi-layer perceptron nerve networks (MLP), conditional variance heterogeneity models (ARCH) and self-regression model and conditional variance (GARCH) (P, Q) have been used. The statistical population of the study is the Tehran Stock Exchange index for the period of April 2008 to April 2018 . The research seeks to reject or confirm the hypothesis that "the use of an artificial neural network and conditional variance models increases the accuracy of the forecast of stock market fluctuations in the Tehran Stock Exchange relative to the conditional variance model" . The results, confirm the validity of the above hypothesis.