Loading...

Optimization of Foreign Exchange (Forex) Trading Using Machine Learning Methods

Fakoor, Mohammad Mahdi | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58410 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Hassan Nayebi, Erfan
  7. Abstract:
  8. The foreign exchange market, commonly known as Forex, is one of the largest and most significant financial markets in the world, attracting the attention of numerous investors on a daily basis. One of the main challenges faced by traders in this market is the accurate prediction of currency prices. Although Forex market forecasting is highly popular, the inherent complexity of this market continues to make accurate prediction a persistent concern. In recent decades, remarkable advancements have occurred in the field of machine learning, particularly in deep learning. These developments have also influenced the Forex market, resulting in the publication of numerous research articles aimed at improving forecasting accuracy. One of the widely used tools in this domain is Long Short-Term Memory (LSTM) neural networks. Specifically designed for time series data analysis, LSTM networks have been employed in Forex trend prediction due to their exceptional ability to learn complex patterns. Nevertheless, there is still a need for a comprehensive and standardized approach to effectively utilize LSTM networks in Forex forecasting. The aim of this research is to investigate the capabilities of deep learning—particularly LSTM networks—in predicting Forex market trends and to enhance prediction accuracy through improved LSTM algorithms. In this study, relevant technical indicators related to Forex trading were thoroughly collected and analyzed. Effective features were then selected as inputs to the model. Using these inputs, an LSTM-based neural network was developed to enable more accurate price trend predictions, providing a potentially valuable tool for traders. A key innovation of this research is the introduction of a novel validation method designed to overcome the limitations of traditional time series analysis techniques. Additionally, a new LSTM algorithm is proposed that utilizes hybrid activation functions in its hidden layers. Experimental results demonstrate that this algorithm significantly outperforms standard LSTM models. For instance, the proposed approach achieves a 1.02% improvement in prediction accuracy. Finally, comparative analyses were conducted between the proposed LSTM model and other deep learning models, including Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs). The results indicate that the proposed LSTM algorithm offers superior performance. This study aims to make a meaningful contribution to the development of financial forecasting methods and to provide practical solutions for traders
  9. Keywords:
  10. Foreign Exchange (Forex) ; Machine Learning ; Artificial Intelligence ; Neural Network ; Exchange Market ; Long Short Term Memory (LSTM) ; Exchange Rate Growth

 Digital Object List

 Bookmark

No TOC