Predicting Stock Exchange Closing Prices using Machine Learning Techniques

Abstract

Predictive analysis using machine learning is becoming more popular, especially when using historical data. Forecasting closing prices in the stock market is intricate owing to the non-linear and unpredictable attributes of financial data. To address these complexities, we propose a customized approach in this study that integrates two distinct models: Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM).Using historical data, our suggested method seeks to forecast stock exchange closing prices. Amreli Steel Limited (ASTL), Meezan Bank Limited (MEZL), United Bank Limited (UBL), and Oil Gas Development (OGDC) are the four corporations whose stock exchange data is included in the dataset. Financial information such as Open, High, Low, Average, and Last Day Close Price (LDCP) are independent variables utilized as inputs for the models. Our primary goal in this research is to assess the efficiency and precision of the introduced models in forecasting closing prices within the stock exchange. Our emphasis is on conducting a comprehensive evaluation of both the ANN and LSTM models in this particular scenario, employing thorough testing and analytical procedures. The results of our research will help the stock market grow by offering useful tips for forecasting stock values. Investors and other market participants can make well-informed decisions and reduce risks with the help of accurate and trustworthy projections. It is crucial to emphasize that our suggested algorithm incorporates cutting-edge machine learning models, going beyond conventional approaches. These models may be able to identify intricate dynamics and patterns in the financial data, producing forecasts that are more precise. Additionally, the inclusion of numerous independent variables enables a thorough examination of the many aspects that affect stock values