زهرا ریحانی یام

In the financial literature, there is much interest in stock price forecasting. Stock price forecasting has become a necessity from the past until now, and this makes us always look for optimal and highaccuracy models for stock price forecasting. In finance, many phenomena are modeled as time series? For this reason, time series models are used to forecast stock prices. With the emergence of artificial intelligence algorithms, a new evolution has been created in time series prediction models for stock price prediction and these models were gradually used to predict stock prices. In this thesis, we try to compare classic time series models, such as exponential smoothing models and Autoregressive integrated moving average, with recurrent neural network models, such as Long Short-Term Memory models and Gated recurrent unit in stock price forecasting. For this purpose, we have implemented the mentioned models on the data of the Tehran Price Index and finally we have done a comparison between the mean square error. The obtained results confirm the significant reduction of prediction error when applying the gated return unit model.