Quantitative researcher working at the intersection of econometrics, machine learning, and financial risk. MSc in Quantitative Finance (Erasmus University Rotterdam); BSc in Econometrics & Data Science (University of Amsterdam), with an exchange at Peking University. I like volatility modelling, forecast evaluation done honestly, and turning research into code that actually runs.
Interests: realized-volatility forecasting · online / sequential learning · e-values and decisions under uncertainty · tail risk and Growth-at-Risk · GARCH-family models · portfolio construction · quantile methods.
- shape-of-evidence — my MSc thesis: how a statistician should design e-value evidence when the decision-maker's loss is known. Derives the unique loss-optimal profile, shows when the likelihood ratio leads to the wrong action, and applies it to Polymarket order-book data on the 2026 World Cup final.
- online-learning-rates-volatility — Model-Assisted Online Learning (MAOL): a parameter-free three-layer online framework for rates-volatility forecasting and VaR calibration, with proved finite-time regret and calibration bounds. Evaluated on 40 years of US Treasuries plus Bund/Gilt/JGB cross-country validation.
- growth-at-risk-replication — from-scratch Python reimplementation of Brownlees & Souza's Growth-at-Risk pipeline: quantile regression, panel QR, and GARCH-based conditional densities across 24 economies, with tick-loss / DQ / DM evaluation.
- garch-volatility-spillovers — do market volatility spillovers improve stock-level GARCH forecasts and the portfolios built from them? Variance-targeting MLE (C++ inner loops), rolling OOS forecasting, and minimum-variance CAPM portfolios on S&P 500 constituents.
- rv-forecasting-har-nn — my BSc thesis (8.1/10): an honest out-of-sample test of neural networks vs. HAR-RV and GARCH for realized volatility. Spoiler: the parsimonious HAR-RV wins.
- vwce-forecasting — self-directed project: GJR-GARCH + jump + Bayesian-shrinkage Monte Carlo of long-horizon ETF wealth, plus production automated-DCA execution through Interactive Brokers.
Python · R · C++ (Rcpp) · STATA · LaTeX · GARCH/HAR · quantile regression ·
neural nets · Monte Carlo · time-series & forecast evaluation
Based in Athens · open to quantitative research roles and collaboration.