Supervisor
Kashif Qureshi
Programme
MSc in Data Analytics
Subject
Computer Science
Abstract
Cryptocurrency markets pose analytical challenges due to their volatility, non-stationarity, and complex interdependencies. Traditional econometric models have struggled with these features, while machine learning models often face overfitting and limited interpretability. This thesis has developed an integrated framework combining clustering, lead–lag detection, and predictive modelling to examine whether structural and temporal dependencies in cryptocurrency markets can improve forecasting performance.
Hourly OHLCV data for a selection of liquid cryptocurrencies from the Binance API over a thirteen-month period has been analysed. Dynamic Time Warping (DTW) has been used to identify asset clusters, while cross-correlation, Granger causality, and Vector Autoregression (VAR) have been tested for leader–lagger dynamics. Forecasting models, including Persistence benchmarks, Ridge regression, ARIMAX, and Long Short-Term Memory (LSTM) networks, have been evaluated for predictive accuracy with and without leader signals.
The results have shown that clustering yields largely homogeneous groupings, reflecting the synchronous nature of cryptocurrency markets. Lead–lag analysis has revealed occasional leader–follower relationships, though these have been unstable and regime-dependent. Predictive modelling has indicated that Ridge regression offers negligible gains, and while LSTM networks have captured some non-linearities, overall improvements over baselines remain modest. Leader signals have not consistently enhanced forecasting accuracy.
The study’s contribution lies in presenting a reproducible analytical pipeline that integrates descriptive and predictive approaches within a single framework. While predictive gains have been limited, the framework offers a foundation for future research incorporating macroeconomic indicators, sentiment analysis, or regime-switching models.
Date of Award
2025
Full Publication Date
2025
Access Rights
open access
Document Type
Capstone Project
Resource Type
thesis
Recommended Citation
Varga, V.
(2025) Clustering and Predictive Modelling of Cryptocurrencies: An Empirical Study of Lead–lag Dynamics and Forecasting Performance. CCT College Dublin.
DOI: https://doi.org/10.63227/652.299.85