An Integrated Approach Towards Stock Price Prediction using LSTM Algorithm

Abstract

Investing in stock markets is always a bit complex and unpredictable for numerous reasons. This study is made by an extensive study of various machine learning and deep learning algorithms which would help reducing the risk in predicting the stock prices. As part of the study, Tehran Stock Exchange is considered in the sectors of non-metallic minerals, basic metals, and finance for testing new ideas. Various algorithms such as extreme Gradient Boost (XGBoost), Support Vector Machine, Random Forest, Decision Tree, K-Nearest Neighbors, Naive Bayes, Logistic Regression and Artificial Neural Network (ANN) are explored, compared, and analyzed in this work. Apart from the above-mentioned algorithms, this research study has used two powerful deep learning techniques, they are Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM). This research study will implement around ten technical indicators that have been gathered over a decade. Stock trading values are utilized to calculate the indicators whereas, binary data is used to convert the indicators. Each prediction model is evaluated by three metrics based on the input way. For continuous data, the evaluation results depict that RNN, and LSTM overcome other prediction models to a greater extent. These deep learning are equally good at evaluating binary data, but this difference hinders due to a considerable enhancement in the model’s performance.