Predicting Stock Price and Risk Using a Custom-Built GRU-Based Application
Predicting Stock Price and Risk Using a Custom-Built GRU-Based Application

Summary/Abstract: Price fluctuations and the loss risk are major problems in stock investing that need to be managed quickly and efficiently. The use of technology can improve the efficiency of price prediction and loss risk assessment. This study aims to implement the Gated Recurrent Unit (GRU) and Value-at-Risk (VaR) algorithms for price prediction and loss risk, packaged in a GUI-based application. GRU is designed to efficiently process sequential data by addressing the problem of short-term memory. VaR is a loss prediction method based on historical returns and quantiles of their distributions. There are two novelties in this study: first, the development of a hybrid model that integrates the GRU and VaR models into a single prediction model. The GRU prediction results are used as input values for the VaR model. Second, the development of a GUI-based application to accelerate prediction results. The results indicate that integrating GRU with VaR provides accurate results. This finding refers to the accuracy of the value and its conformity with the actual data. The GUI application consists of four main menus: data input, preprocessing, price prediction, and loss prediction. The GUI application has been proven to improve the prediction process without reducing accuracy.

• Details
• Contents