Foreign exchange prediction and trading using feedforward neural networks

• Main Author: Han, Hongyi
• Other Authors: Wang Lipo
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78029
• Institution: Nanyang Technological University

The foreign exchange (forex) market is one of the most widely traded markets in the world, with an average daily transaction volume exceeding US$5 trillion. It is one of the world's largest and most liquid financial markets, with different currencies being exchanged continuously where individuals, companies and organizations conduct global business 24 hours a day from Sunday evening to Friday night. Forex trading allows traders to take advantage of exchange rate fluctuations in large foreign exchange currency pairs, and speculate that the price of a country's currency will rise or fall relative to the price of another country's currency, thus selling or buying the pair. Hence it is universally acknowledged that an intelligent data analysis system for prediction of forex is of vital importance for traders in decision making. Five different Artificial Neural Network architecture models such as Convolutional Neural Network (CNN), Multi-layered Perceptron Neural Network (MLP), Simple Recurrent Neural Network (SRNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) are used to predict the foreign exchange rates of European Euro against US Dollar (EUR/USD), US Dollar against Japanese Yen (USD/JPY), and British Pound Sterling against US Dollar (GBP/USD). The project consists of two parts. The first part is to reproduce the work by a published research paper which used CNN to predict the forex and to establish benchmark outcomes. The second part aims to improve the prediction capability, that constructs different models such as MLP, SRNN, LSTM and GRU to enhance the performance compared to the benchmark result. The project will be implemented using the open source neural network library Keras using Tensorflow as the backend.