Forecasting Stocks Prices with GRU and Attention Mechanism

Stocks represent a type of investment, reflecting ownership of a company. However, investing in stocks carries risks due to the potential for changes in share values. AI has gained significant traction in the financial sector, particularly in stock markets. To reduce mistakes in investing in shares, you can use Artificial Intelligence (AI). This capability has proven invaluable in forecasting stock prices, but the main challenge lies in minimizing the error rate of the algorithms used. Additionally, there is a demand for lightweight machine-learning architectures capable of fast and efficient processing. This research aims to overcome this problem by utilizing the GRU algorithm and the attention gate mechanism to design an architecture that is lighter than existing research while increasing the performance of the forecasting stocks value. The results of the proposed model provide better results when compared to previously existing architectures, where the proposed model obtains RSME results of 0.0243.