Quantitative research · Machine learning · Data science
I explore financial data through statistical modelling, machine learning, and numerical methods. My projects span equity factors, central-bank communication, interpretable credit scoring, and market microstructure.
LinkedIn · Research reports · PyData Global 2025 talk
I teach Applied Probability & Quantitative Reasoning to graduate students in the Master 2 Finance Technology Data.
Random Walks, Markov Chains & Gambler's Ruin — lecture and exercises
A course built around probabilistic reasoning: choosing the Markov state, deriving ruin probabilities and expected stopping times, solving linear recurrences, and connecting analytical results to simulation.
- Lecture slides and worked problem sets, including Gaussian random walks and waiting times for consecutive outcomes.
- Expanded notes with derivations, recurrence and return-time distinctions, and exercises with solutions.
- A reproducible Python lab comparing exact results with Monte Carlo simulation.
| Project | Research question and methods | Format |
|---|---|---|
| European equity factor research | Factor construction, cross-sectional neutralization, information coefficients, and portfolio diagnostics. | Python research pipeline; committed charts are in-sample. |
| Central-bank communication and market reactions | Text similarity, financial sentiment, and econometric analysis of ECB communication; includes BERT and VADER experiments. | Replication and extension, notebooks, and report. |
| Explainable credit scoring | Decision-tree rules, interaction features, and loan-risk segmentation by count and exposure. | Exploratory notebook. |
| Order-book sequence classification | Stock identification from order-book events using bidirectional GRUs and a CNN–GRU variant. | Challenge experiments; validation remains to be strengthened. |
| Trust dynamics in Bitcoin networks | Network analysis, anomaly detection, and modelling of trust dynamics. | Collaborative academic project with Gwendoline Hays-Valentin. |
| Optimal binning in logistic regression | Discretization, Weight of Evidence, interpretability, and model evaluation. | PyData Global 2025 presentation. |
- Quantitative research: factor models, financial econometrics, time series, simulation, and numerical methods.
- Machine learning: tabular models, sequence models, NLP, and interpretable modelling.
- Data science: Python, pandas, Polars, NumPy, scikit-learn, TensorFlow/Keras, and research visualization.
- Numerical programming: C++ learning exercises, including Monte Carlo integration and basic pricing examples.
Repositories include research prototypes, coursework, and learning resources. The selection above provides a starting point; individual project READMEs describe their scope and current limitations.