Modeling the return fluctuationsTime series and artificial neural network

In this research, we modeled the yield fluctuations of the Tehran Stock Exchange index. For thispurpose, in the linear method section, predicting the yield of basic metals index, from two methods ofsingle-variable time series that do not consider long-term memory and time series that consider long-term memory is done. Since this research, on the one hand, intends to model the fluctuations of TehranStock Exchange index, comparing two methods of time series and artificial neural network, and thepurpose of the research was directed towards the scientific application of this knowledge, so theresearch is practical according to the purpose, and on the other hand, As it intends to investigate thechange process of a specific variable (stock market index) over time, based on the nature and methodof the research, it is considered a survey descriptive research. In order to predict and examine theresults of using artificial neural network models, the GARCH model was used to estimate the TehranStock Exchange index to reflect the effect of clustering of fluctuations in order to improve theprediction results. In order to check the hypotheses and answer the research questions, the time seriesmethod and the artificial neural network method were used. The statistical population sampled in thisresearch is the companies accepted in the Tehran Stock Exchange, which was used with the methodof systematic removal and application of restrictions of basic metal companies active in the TehranStock Exchange in the period of ۱۳۹۷ to ۱۴۰۱. Finally, according to the confirmation of the hypotheses,the results were confirmed. The combined models significantly reduce the intra-sample predictionerror.

Keywords:

Efficiency Fluctuations , Time Series , Artificial Neural Network , GARCH model