Forecasting Currency Exchange Trend on USD/CAD
Akubue, Simon (2024)
Akubue, Simon
2024
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2024122238069
https://urn.fi/URN:NBN:fi:amk-2024122238069
Tiivistelmä
Currency exchange rate forecast has been, and remains a challenging tasks. With unpredictable natural disasters, political instabilities, government policies, and many other factors, it becomes difficult to correctly forecast the currency exchange rate. Many researchers in the past have done great works on forecasting the exchange rate of the United State Dollar (USD)/Canadian Dollar (CAD) using statistical approach. Even the Fundamental approach of relying on macroeconomic factors of the two countries, such as GDP ratio, Import/Export, government revenue, etc. were considered at various points. But (while forecasting the USD/CAD exchange rate), none of the previous methods considered deeply the underlying market trends which forms the basics of Technical analysis. We have included various machine learning models and time sensitive indicators that directly aligns with the USD/CAD exchange rate movement so as to improve on this issue. These features will create a new dimension for researchers to predict and forecast the USD/CAD exchange rate. We have considered various types of models for predicting and forecasting the USD/CAD exchange rate, and realized that among all our models, Time Series models provides the best accuracy. While building our model, I relied of daily forex statistical data maintained by Tick Data Suite (TDS) and Investing.Com. All our macroeconomics data (Consumer Price Indices (CPI), Interest Rates, Imports, Exports, and Un Employment Rates) were automatically downloaded from the Federal Reserve Economic Data (FRED)’s portal via an API. Due to huge data size available to me, I used 90% of my data for training my models, and the remaining 10% for my testing. At the end, we were able to come up with good results from our models, of which Linear Regression, Bayesian Ridge Regression, and Random Forest Regressor (RFR) outperformed others based on accuracy.