Machine Learning Techniques for Forecasting Foreign Exchange Rates: A State-of-the-Art Review

Thumbnail Image

Files

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Bachelor's thesis
Electronic archive copy is available locally at the Harald Herlin Learning Centre. The staff of Aalto University has access to the electronic bachelor's theses by logging into Aaltodoc with their personal Aalto user ID. Read more about the availability of the bachelor's theses.
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-10-07

Department

Major/Subject

Data Science

Mcode

SCI3095

Degree programme

Aalto Bachelor’s Programme in Science and Technology

Language

en

Pages

34

Series

Abstract

This thesis presents a state-of-the-art review of machine learning techniques for forecasting foreign exchange rates. It examines the models and data types that have been used to forecast foreign exchange rates in 11 journal articles published between 2020 and 2025. These models are categorized into three families: traditional machine learning, deep learning, and hybrid. On the other hand, the data types are technical, fundamental, text, and others. In this collection of journal articles, traditional ML models appear competitive for forecasting in the short-term, due to their efficiency and interpretability. On the other hand, deep learning approaches are highly prominent, due to their ability to capture complex, nonlinear, and noisy information. Despite this, deep learning models can struggle to forecast FX rates at times. Hence, the incorporation of domain knowledge, proper integration of different data types, and hybridization with other techniques have been employed to significantly improve their predictive performance. Despite these advances, the current literature demonstrates little investigation with more recent deep learning architectures, restricted incorporation of different data types, and an emphasis on statistical rather than economic performance metrics. As a result, future research should investigate transformer-based models, further explore multimodal approaches, and evaluate models with real-time trading strategies.

Description

Supervisor

Korpi-Lagg, Maarit

Thesis advisor

Yılmaz, Ersin

Keywords

machine learning, deep learning, hybrid, foreign exchange rates, FX, forecasting

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202510218404

Collections

[kand] Perustieteiden korkeakoulu / SCI