Developed a robust ML pipeline for S&P 500 sector rotation forecasting (2000–2025), leveraging the AutoGluon-Tabular framework to ensemble tree-based algorithms trained on approximately 450 engineered macroeconomic and technical features. Validated through strict Point-in-Time rolling windows with purge gaps, the core research evaluates 14 risk-filtering mechanisms across multiple time horizons using VectorBT, Alphalens, and QuantStats. Results demonstrate that predictive accuracy (Information Coefficient) becomes statistically significant at the quarterly horizon, while regime-based filters prove essential for strategy viability; successfully reducing maximum drawdown by 73% and generating positive Alpha compared to the high-risk, negative-Alpha profile of unfiltered ML predictions.

Co-authored with another member.

Analysis of how decentralized finance is reshaping market microstructure, structured in two parts. Part I covers the shift from Central Limit Order Books to Automated Market Makers, explaining the constant product formula (x·y=k) that governs liquidity pools, the "lazy liquidity" inefficiency of uniform distribution across all price ranges, and Uniswap v3's concentrated liquidity as a solution. Part II examines the "dark forest" of transaction ordering: how the public mempool creates an adversarial pre-execution environment, the MEV extraction pipeline, and a quantitative framework for measuring MEV as a component of total execution cost. Empirical data on sandwich attacks, high attacker concentration, and the distinction between price efficiency through arbitrage and execution fairness for end users.

Co-authored with four students from three Greek universities.

Comprehensive analysis of Jumbo A.E.E., one of Greece's largest retail chains, produced as part of SED's Listed Companies General Meetings Monitoring Programme. The report covers the company's corporate profile and historical expansion, with detailed financial analysis including revenue trajectory, profitability metrics, stock performance and market capitalization, seasonality patterns in sales, the COVID-19 impact and subsequent recovery, competitive positioning within Greek retail, capital structure events (IPO, bond issuances, share capital increases), board composition and governance structure, and CSR activities. The analysis concludes with coverage of the 2023 Annual General Meeting.

Recognized for achieving the highest overall performance among peers, evaluated against a comprehensive set of quantitative and qualitative KPIs. My team secured the #1 national ranking for three out of five months and maintained a top-3 standing nationally throughout the semester.

Our project was selected via a competitive voting process to be presented at Pfizer's Center for Digital Innovation (CDI) in Thessaloniki.