Advanced AI and Machine Learning Models in Cryptocurrency Market Analysis and Prediction - Dissertation Thesis

Aug 4, 2026

Upload by Riley

This dissertation investigates the efficacy of advanced AI and machine learning models for Bitcoin price prediction, tackling the cryptocurrency market’s volatility and structural complexity. The methodology integrates heterogeneous data sources, including technical indicators, macroeconomic factors, and social media analysis, to improve forecasting accuracy. A set of time-series approaches is developed and evaluated: statistical models such as SARIMAX, ML models including Prophet, and deep learning methods like DeepAR and Temporal Fusion Transformers. The study proposes a novel AI multi-agent framework with dedicated agents for financial analysis, research, and investment recommendations, enriching predictions with contextual insights.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages69

Reading Time2 h 54 min

AccessFree to download

Introduction

Background and Motivation

Research Objectives

Significance of the Study

Literature Review

Cryptocurrency Market Analysis

Machine Learning in Financial Forecasting

Multi-Agent AI Systems in Finance

Methodology

Scope of the study

Data Collection and Preprocessing

Feature Selection

Time-Series Models for Bitcoin Price Prediction

Stationarity Tests

Multi-Agent Framework

Results and Analysis

Model Performance Evaluation

Multi-Agent System Output Analysis

Conclusion

Implications for Cryptocurrency Market Analysis

Limitations of the Study

Future Research Directions

Appendix

Technical Implementation Details

Data Dictionary

References

INTERDEPARTMENTAL PROGRAMME OF POSTGRADUATE STUDIES IN BUSINESS ADMINISTRATION Dissertation Thesis Advanced AI and Machine Learning Models in Cryptocurrency Market Analysis and Prediction by Stamatis Kavidopoulos Under the supervision of: Theodore Panagiotidis, Professor Thesis submitted for the degree of Master in Business Administration September, 2024 Acknowledgments I would like to express my deepest gratitude to several individuals who have been instrumental in the completion of this dissertation. First and foremost, I extend my heartfelt thanks to my advisor, Professor Theodore Panagiotidis, for his unwavering guidance and support throughout this academic journey. His insightful feedback and encouragement have been invaluable in shaping both this research and my growth as a scholar. I am eternally grateful to my wife Aphrodite, for her boundless love, patience, and understanding during the countless hours devoted to this work. Her unwavering belief in me and her constant support have been my anchor throughout this challenging process. To my beloved daughter Marilia, whose innocent curiosity and joy have been a constant source of inspiration and motivation. The sacrificed playtime and bedtime stories will forever be appreciated, and I hope that one day this work will make you proud. This achievement would not have been possible without the love and support of my family. Your encouragement has been the driving force behind my perseverance. Thank you all for being an integral part of this academic milestone in my life. Abstract This dissertation investigates the efficacy of advanced AI and machine learning models for Bitcoin price prediction, addressing the inherent volatility and complexity of the cryptocurrency market. The study employs a multi-faceted approach, integrating diverse data sources, including technical indicators, macroeconomic factors, social media analysis, to enhance predictive accuracy. We develop and evaluate various time-series models, including statistical models like SARIMAX, machine learning models such as Prophet, and deep learning architectures like DeepAR and Temporal Fusion Transformers (TFT) . Furthermore, we introduce a novel AI multi-agent framework comprising specialized AI agents for financial analysis, research, and investment recommendations. This framework enriches the prediction models with contextual insights and facilitates a more comprehensive understanding of market dynamics. Our results demonstrate the superior performance of Prophet, in capturing both long-term trends and short-term fluctuations of Bitcoin price movements. The integration of macroeconomic and technical indicators further enhances predictive accuracy, highlighting the importance of incorporating market context. The developed multi-agent framework showcases the potential of combining specialized AI agents for cryptocurrency market analysis, enabling more robust and well-informed investment recommendations. This research contributes to the expanding field of financial forecasting using AI, providing valuable insights for traders, investors, and researchers navigating the dynamic cryptocurrency landscape. Keywords Bitcoin, Cryptocurrency, Price Prediction, Time Series Analysis, Deep Learning, AI Multi-Agent Systems Disclaimer The contents of this dissertation, including all analyses, frameworks, and models presented, are intended for academic purposes only. The Multi-Agent Framework Architecture and its associated BTC forecasting model are designed to demonstrate the application of multi-agent systems and machine learning techniques in financial analysis. It is important to note that the information and results provided herein do not constitute financial advice. The author and associated institutions do not endorse any investment decisions or strategies based on the findings of this research. Readers are strongly advised to consult with a certified financial advis

FAQs

What is the dissertation’s main goal for cryptocurrency forecasting?

To assess how advanced AI and machine learning models perform in Bitcoin price prediction while addressing market volatility and complexity.

Which types of prediction models does the study evaluate?

It evaluates statistical models (e.g., SARIMAX), machine learning models (e.g., Prophet), and deep learning architectures (e.g., DeepAR and Temporal Fusion Transformers).

How does the multi-agent framework contribute to the results?

The framework uses specialized agents for financial analysis, research, and investment recommendations, adding contextual insights to strengthen understanding and improve recommendation quality.

Aug 4, 2026

Upload by Riley