Cryptocurrencies’ Prices Discovery Through Machine Learning Algorithms : Bitcoin and Beyond
Islam, Mominul (2023)
2023
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20231123148589
https://urn.fi/URN:NBN:fi-fe20231123148589
Tiivistelmä
The evolution of cryptocurrencies has emerged as a fundamental shift in the financial landscape, with price discovery being an area of intense interest and complexity. The thesis titled “Cryptocurrencies’ price discovery through machine learning algorithms: Bitcoin and beyond” aims to investigate and unravel this complexity through the lens of machine learning.
In this comprehensive study, four major machine learning algorithms - Logistic Regression (LR), Decision Tree, Random Forest (RF), and Support Vector Machine (SVM) were applied to forecast the daily prices of four leading cryptocurrencies: Bitcoin, Ethereum, Cardano, and Solana, alongside an analysis of hourly Bitcoin price prediction.
The findings reveal distinct performance characteristics for each algorithm. Logistic Regression exhibited high accuracies for Bitcoin and Ethereum daily predictions at 0.86 and 0.85, respectively. Support Vector Machine proved particularly effective for Cardano and Solana with accuracies of 0.90 and 0.97. Conversely, the Decision Tree and RF algorithms demonstrated more modest performance across the examined cryptocurrencies. Besides, a specialized investigation into Bitcoin’s hourly price prediction, employing the same set of algorithms, yielded varying results, with LR showing a standout accuracy of 0.98.
This research encompasses a journey from the foundational principles of cryptocurrency to the advanced techniques of machine learning, highlighting both the opportunities and challenges inherent in this rapidly evolving field. It acts as a roadmap for future investigations, offering the potential to deepen our understanding of cryptocurrencies’ impact on the global financial landscape and to extend the boundaries of knowledge in the area of price discovery through machine learning.
In this comprehensive study, four major machine learning algorithms - Logistic Regression (LR), Decision Tree, Random Forest (RF), and Support Vector Machine (SVM) were applied to forecast the daily prices of four leading cryptocurrencies: Bitcoin, Ethereum, Cardano, and Solana, alongside an analysis of hourly Bitcoin price prediction.
The findings reveal distinct performance characteristics for each algorithm. Logistic Regression exhibited high accuracies for Bitcoin and Ethereum daily predictions at 0.86 and 0.85, respectively. Support Vector Machine proved particularly effective for Cardano and Solana with accuracies of 0.90 and 0.97. Conversely, the Decision Tree and RF algorithms demonstrated more modest performance across the examined cryptocurrencies. Besides, a specialized investigation into Bitcoin’s hourly price prediction, employing the same set of algorithms, yielded varying results, with LR showing a standout accuracy of 0.98.
This research encompasses a journey from the foundational principles of cryptocurrency to the advanced techniques of machine learning, highlighting both the opportunities and challenges inherent in this rapidly evolving field. It acts as a roadmap for future investigations, offering the potential to deepen our understanding of cryptocurrencies’ impact on the global financial landscape and to extend the boundaries of knowledge in the area of price discovery through machine learning.
Kokoelmat
- 512 Liiketaloustiede [502]