Tsioumpris Ioannis (Phd Candidate)

Thesis title: Time Series Analysis with Machine Learning Methods

Supervisor: Sidiropoulos Antonis

Advisory Committee Members:

Antoniou Efstathios, Professor, Dept. of Information and Electronic Engineering, IHU
Ougiaroglou Stefanos, Assistant Professor, Dept. of Information and Electronic Engineering, IHU

Abstract:

This research investigates the correlation between multiple interrelated time series, focusing on how short-term price fluctuations in one series can influence another. The study primarily targets the Forex and cryptocurrency markets, examining the extent to which significant price changes in one currency pair (e.g., EUR/USD) impact other pairs (e.g., AUD/CZK) within short time windows.

The research employs advanced time series analysis techniques and machine learning models, such as LSTM (Long Short-Term Memory), XGBoost, Random Forest, and VAR (Vector Autoregression), to uncover patterns and predict short-term price changes. High-frequency data from Forex and cryptocurrency markets serve as the primary dataset, highlighting both direct and indirect correlations between pairs.

By leveraging methodologies like correlation analysis, volatility examination, and cointegration tests, the research aims to model and quantify the dynamic relationships between time series. The findings are expected to contribute to the field of algorithmic trading by providing actionable insights for traders and analysts seeking to exploit price movements in interconnected markets.

The outcomes of this study are anticipated to extend beyond financial markets, offering applications to other domains with similar time series behaviors, such as weather forecasting.