Presenting a hybrid model for predicting stock market index volatility an‎d estimating value at risk using developed GARCH family models an‎d deep learning (case study of Tehran Stock Exchange Index)

1/1/1900 12:00:00 AM

Investment means choosing assets to keep an‎d earn more income in order to gain profit an‎d better economic prosperity in the future. Considering the limited resources of investment in the country, economists an‎d financial specialists are always looking for ways to make the best use of economic resources to provide investors. In the meantime, the stock market plays an essential role in allocating capital resources to productive sectors by attracting small an‎d stray resources. The transparent structure an‎d framework of the stock market in pricing an‎d disseminating information related to the financial value of economic enterprises has led to the recognition of this market as a representative of the real state of the economy. Therefore, the ability of the stock market to quickly an‎d accurately reflect all available information has always been the focus of financial analysts, legislators an‎d economists. The stock market is driven by the emotions of those who participate in it, as their desire to buy a particular stock creates deman‎d, an‎d their desire to sell a stock creates supply. These dynamics of supply an‎d deman‎d still occur in the stock market. Therefore, people are always looking for ways to reduce risk, because risk tolerance in the stock market is affected by volatility. As a result, it is necessary to study market fluctuations before making any investment. Volatility is an indicator of unexpected changes in asset returns in a period of time. It is the degree of fluctuations in the price of financial assets an‎d the uncertainty criterion of asset returns, which is used to measure the risk of financial assets. The higher this fluctuation is, the higher the range of fluctuations in the price of financial assets, an‎d the uncertainty in the returns on assets increases. In addition to assessing the risk of financial assets, volatility is not only an important parameter for portfolio selec‎tion, risk management, monetary policies an‎d pricing, but it is also an important factor in determining different value-at-risk models that calculate the market risk by calculating the maximum loss that can be incurred in a The target period is measured according to the level of confidence in normal market conditions. In this research, at first, using the developed models of the GARCH family, we predict the returns an‎d fluctuations of the Tehran Stock Exchange index. In the next step, the results obtained from the developed classic time series models are entered into a learning LSTM algorithm as initial values. They are deepened an‎d finally, by means of a combined model of volatility an‎d value, they are exposed to the risk of the stock market index. Keywords: volatility prediction, value at risk, deep learning LSTM algorithm, extended autoregressive conditional heteroskedasticity variance models family, stock market index

پيش‌بيني نوسانات , ارزش در معرض ريسك , الگوريتم حافظه كوتاه مدت طولاني , مدل‌هاي واريانس ناهمساني شرطي خودرگرسيوني تعميم يافته , شاخص بورس

volatility prediction , value at risk , deep learning LSTM algorithm , extended autoregressive conditional heteroskedasticity variance models family , stock market index

Fatemeh Mesbahi

Dr. Seyed Jafar Sajadi