Prediction of Bitcoin Daily Returns Based on OLS, XGBoost, and CNN Machine Learning Models

Ye He

2024

Abstract

This study seeks to predict Bitcoin's daily return through a comparison of three machine learning models: Ordinary Least Squares (OLS), XGBoost (Extreme Gradient Boosting), and Convolutional Neural Network (CNN). To assess the effectiveness of these models in capturing Bitcoin market fluctuations, the relevant market data is first cleaned and standardized, followed by training and testing with the three models. The findings reveal that the OLS model excels in stable market conditions, exhibiting a smaller prediction error. Meanwhile, the XGBoost model shows promise in handling nonlinear relationships and market fluctuations, albeit with a larger prediction error. Unfortunately, the CNN model did not meet expectations, struggling to effectively capture the market's complex characteristics. According to the analysis, this research highlights that various machine learning models demonstrate differing applicability for predicting Bitcoin returns across diverse market environments. Future studies could enhance prediction accuracy by optimizing model parameters and incorporating additional feature variables.

Paper Citation

in Harvard Style

He Y. (2024). Prediction of Bitcoin Daily Returns Based on OLS, XGBoost, and CNN Machine Learning Models. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 181-187. DOI: 10.5220/0013212400004568

in Bibtex Style

@conference{ecai24,
author={Ye He},
title={Prediction of Bitcoin Daily Returns Based on OLS, XGBoost, and CNN Machine Learning Models},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={181-187},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013212400004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Prediction of Bitcoin Daily Returns Based on OLS, XGBoost, and CNN Machine Learning Models
SN - 978-989-758-726-9
AU - He Y.
PY - 2024
SP - 181
EP - 187
DO - 10.5220/0013212400004568
PB - SciTePress