Forecasting stock prices using deep learning models with technical indicators

Accurate stock price prediction is crucial for supporting investors' decisions regarding the timing and allocation of investments. However, the dynamic, non-linear, complex, and chaotic nature of the stock market makes price prediction a challenging task. Market movements are influenced by various macroeconomic factors, such as political events, corporate policies, economic conditions, commodity prices, and bank rates. Furthermore, advancements in technology and communication systems enable these events to be processed rapidly, leading to swift fluctuations in stock prices. Consequently, banks, financial institutions, and large investors are compelled to make quick buy and sell decisions, further complicating precise predictions. Therefore, there is a pressing need for the development of new and effective methods for accurate stock price predictions. This study aims to predict stock prices using technical indicators obtained from sources like Yahoo Finance. To reduce the noise in raw data, obtain meaningful results and increase the estimation accuracy, correlation coefficient method was used for technical indicators. Artificial Neural Networks (ANN) models provide great efficiency in the analysis of financial time series data. The most appropriate parameters were selected for the estimations to be made with ANN. After the data preprocessing process, Single Layer Long Short Term Memory (LSTM), 3 Layer LSTM, 3 Layer Bidirectional Long Short Term Memory (BiLSTM) and Hybrid Convolutional Neural Network-Long Short Term Memory (CNN-LSTM); methods were applied. With this model, in addition to finding realistic price estimations, it was aimed to reduce the features affecting stock price estimation through technical indicators. The results showed that the Single Layer LSTM method provides more realistic estimations compared to other methods among deep learning (DL) techniques.