Foreign exchange rate prediction using feedforward neural networks
The foreign exchange (forex) market concerns everyone. From governments trying to build a stronger currency to have an edge in international trading to hedge funds using algorithmic trading to make a profit, everyone is a part of the forex market. Even an individual exchanging local currency for cur...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/74969 |
| _version_ | 1811683630703968256 |
|---|---|
| author | Khetwani Akash Manish |
| author2 | Wang Lipo |
| author_facet | Wang Lipo Khetwani Akash Manish |
| author_sort | Khetwani Akash Manish |
| collection | NTU |
| description | The foreign exchange (forex) market concerns everyone. From governments trying to build a stronger currency to have an edge in international trading to hedge funds using algorithmic trading to make a profit, everyone is a part of the forex market. Even an individual exchanging local currency for currency of his vacation destination is participating in this market. Thus, it is no surprise that researchers have been trying to predict forex rate movement since a long time. This project an attempt to make a positive contribution to this research topic by using a variation of a Feedforward Neural Network to predict exchange rate. A Random Vector Functional Link (RVFL) Neural Network is one such variation which has not been deeply explored in this domain. The project uses an RVFL network to predict the US Dollar against Indian Rupee (USD/INR) exchange rate. The first part of the project consists of reproducing a published research paper to establish a benchmark result. This is done by building a simple feedforward and recurrent neural network. The dataset is the same as the one in the published research paper. The next part consists of designing a robust RVFL network to improve upon this benchmark result. The neural network functionality of MATLAB software has been used for this purpose. The input variables selected are a combination of fundamental factors affecting exchange rate movement. The performance of the network is evaluated by comparing the Mean Square Error (MSE) between the different networks used. Through the project, the author was successfully able to create an RVFL network that has a better performance than the established benchmark. Lastly, a dialogue on future research directions has been presented. |
| first_indexed | 2024-10-01T04:15:48Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/74969 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T04:15:48Z |
| publishDate | 2018 |
| record_format | dspace |
| spelling | ntu-10356/749692023-07-07T16:29:39Z Foreign exchange rate prediction using feedforward neural networks Khetwani Akash Manish Wang Lipo School of Electrical and Electronic Engineering DRNTU::Engineering The foreign exchange (forex) market concerns everyone. From governments trying to build a stronger currency to have an edge in international trading to hedge funds using algorithmic trading to make a profit, everyone is a part of the forex market. Even an individual exchanging local currency for currency of his vacation destination is participating in this market. Thus, it is no surprise that researchers have been trying to predict forex rate movement since a long time. This project an attempt to make a positive contribution to this research topic by using a variation of a Feedforward Neural Network to predict exchange rate. A Random Vector Functional Link (RVFL) Neural Network is one such variation which has not been deeply explored in this domain. The project uses an RVFL network to predict the US Dollar against Indian Rupee (USD/INR) exchange rate. The first part of the project consists of reproducing a published research paper to establish a benchmark result. This is done by building a simple feedforward and recurrent neural network. The dataset is the same as the one in the published research paper. The next part consists of designing a robust RVFL network to improve upon this benchmark result. The neural network functionality of MATLAB software has been used for this purpose. The input variables selected are a combination of fundamental factors affecting exchange rate movement. The performance of the network is evaluated by comparing the Mean Square Error (MSE) between the different networks used. Through the project, the author was successfully able to create an RVFL network that has a better performance than the established benchmark. Lastly, a dialogue on future research directions has been presented. Bachelor of Engineering 2018-05-25T06:35:53Z 2018-05-25T06:35:53Z 2018 Final Year Project (FYP) http://hdl.handle.net/10356/74969 en Nanyang Technological University 59 p. application/pdf |
| spellingShingle | DRNTU::Engineering Khetwani Akash Manish Foreign exchange rate prediction using feedforward neural networks |
| title | Foreign exchange rate prediction using feedforward neural networks |
| title_full | Foreign exchange rate prediction using feedforward neural networks |
| title_fullStr | Foreign exchange rate prediction using feedforward neural networks |
| title_full_unstemmed | Foreign exchange rate prediction using feedforward neural networks |
| title_short | Foreign exchange rate prediction using feedforward neural networks |
| title_sort | foreign exchange rate prediction using feedforward neural networks |
| topic | DRNTU::Engineering |
| url | http://hdl.handle.net/10356/74969 |
| work_keys_str_mv | AT khetwaniakashmanish foreignexchangeratepredictionusingfeedforwardneuralnetworks |