Stock Price Prediction Using Deep Learning Algorithms Based on Technical Indicators

Abstract

Accurately forecasting stock prices helps investors decide when and where to invest. However, the dynamic, non-linear, complex, and chaotic nature of the stock market makes price forecasting difficult. Market movements are influenced by many macroeconomic factors such as political events, corporate policies, economic conditions, commodity prices, and bank interest rates. In addition, advances in technology and communication systems allow these events to be processed quickly, causing stock prices to fluctuate rapidly. Banks, financial institutions, and large investors are therefore forced to act quickly, and the complexity of the market makes accurate forecasting even more difficult. Therefore, new and effective methods must be developed for accurate stock price predictions. In this study, technical indicators were utilized to reduce noise in raw data, obtain meaningful results, and enhance prediction accuracy. Artificial Neural Network (ANN) models demonstrate significant efficiency in analyzing financial time series data. The most appropriate parameters for predictions made with ANN were selected using the correlation coefficient method. The objective was to predict stock prices using technical indicators obtained from websites such as Yahoo Finance. Following data preprocessing, various methods were applied, including Single Layer Long Short-Term Memory (LSTM), 3-Layer LSTM, 3-Layer Bidirectional Long Short-Term Memory (BiLSTM), and Hybrid Convolutional Neural Network-Long ShortTerm Memory (CNN-LSTM). This model aimed not only to find realistic price estimates but also to reduce the features that influence stock price estimates through technical indicators. The results indicate that the Single Layer LSTM method provides more realistic estimates than other deep learning (DL) techniques.