Forecasting the stock market using LSTM neural network hybrid model
رضوان عباسی
1
(
استادیار ،دانشکده مهندسی برق، پزشکی و مکاترونیک، واحد قزوین، دانشگاه آزاد اسلامی، قزوین، ایران
)
طاهره رامه
2
(
گروه فناوری اطلاعات،دانشکده مدیریت و حسابداری، واحد قزوین، دانشگاه آزاد اسلامی، قزوین، ایران
)
محمدرضا ثنایی
3
(
گروه فناوری اطلاعات،دانشکده مدیریت و حسابداری، واحد قزوین، دانشگاه آزاد اسلامی قزوین، قزوین، ایران
)
Keywords: LSTM, Text embedding, Stock marketing,
Abstract :
Forecasting stock prices and returns is one of the most complex and controversial issues in financial markets, which is always the concern of investors and shareholders. The stock market is vulnerable to various factors that affect the price fluctuations in the stock market. The development of a robust stock market algorithm that can accurately predict stock behavior is needed to maximize profits and minimize investor losses. Considering that in addition to the history of each share, psychological factors of the market also affect the value of each share, in this research, a hybrid model of artificial intelligence based on LSTM and text embedding is proposed, which is based on the history of the stock market in the form of data. It pays attention to time series, and it extracts psychological characteristics of the market from news sites and predicts the future of the stock market. The results of the evaluations show that the proposed model can predict the future of the market well and has a lower RMSE than the PROFET and ARIMA methods.