👋 Hi, I’m Thomas

Undergraduate researcher at Sogang University (Seoul, Korea)
Majors in Financial Economics, Econometrics, Big Data Science, Japanese
Minor in AI & Convergence Software Programming

I focus on quantitative finance, causal inference in asset allocation, and financial machine learning.
My recent work centers on building explainable, causality-aware portfolio construction pipelines and robust backtesting frameworks.

---

🔬 Recent Research & Key Projects

Causal Asset Allocation

Causal Inference-Based Portfolio Optimization Pipeline

• Role: Leader / Researcher / Engineer
• Summary:
  A portfolio construction framework that explicitly models causal relationships between macro factors and asset clusters using DAGs. The pipeline supports intervention analysis and scenario-based allocation (e.g., shocks to rates, inflation, volatility).
• Techniques: DAG, Causal Discovery, Double Machine Learning, Intervention Analysis, Portfolio Optimization
• Keywords: Causal Inference, Asset Allocation, Macro Factors, Explainable Allocation

---

Posterior-NCO

Improving NCO Portfolio Allocation via Posterior Covariance and Bayesian Updating

• Role: Leader / Researcher / Engineer
• Summary:
  An extension of Nested Clustered Optimization (NCO) that replaces sample covariance with posterior covariance estimated via Bayesian updating. The framework integrates manager views and uncertainty to improve out-of-sample stability.
• Techniques: NCO, Bayesian Updating, Monte Carlo Backtesting, Portfolio Optimization
• Keywords: Portfolio Construction, Robust Covariance, Bayesian Finance

---

Causal Factor GAN

The Dawn of Explainable Quantitative Finance

• Role: Leader / ML Engineer
• Summary:
  A causal data generating process (DGP) combined with GANs to simulate factor-driven return dynamics under intervention and counterfactual scenarios. Enables stress-testing factor portfolios beyond correlation-based simulations.
• Techniques: DAG-based Causal Modeling, Conditional GAN, QuantGAN, Time Series Generation
• Keywords: Causal Inference, GAN, Factor Investing, Synthetic Data

---

AFML Code Implementation

Reproducible Implementations of Methods from Advances in Financial Machine Learning

• Role: Contributor / Engineer
• Summary:
  A clean, modular Python implementation of core methodologies from Marcos López de Prado’s Advances in Financial Machine Learning, including event-driven labeling, purged k-fold CV, meta-labeling, sample weighting, and fractional differentiation.
• Techniques: Triple Barrier Labeling, Purged K-Fold CV, Meta-Labeling, Sample Weights, Fractional Differentiation
• Keywords: Financial Machine Learning, Reproducibility, Backtesting Infrastructure

---

BOBP ETF Listing & Management

Design, Launch, and Ongoing Management of a Rules-Based ETF

• Role: Product Lead / Quant Researcher
• Summary:
  End-to-end design of the BOBP ETF, including index construction, portfolio rules, rebalancing logic, and live monitoring. Focused on building a transparent, rules-based ETF with robust backtesting and risk controls.
• Techniques: Index Construction, Portfolio Rules, Rebalancing, Risk Management, Live Monitoring
• Keywords: ETF Design, Asset Management, Index Methodology, Portfolio Operations

---

GDP Nowcasting with Explainable Machine Learning

• Role: ML Engineer
• Summary:
  High-frequency macro nowcasting framework using financial and textual indicators. Model explainability is emphasized via SHAP and feature attribution to interpret economic drivers of GDP revisions.
• Techniques: XGBoost, SHAP, Macro Indicators, Text-based Features
• Keywords: Nowcasting, Explainable AI, Macroeconomics

---

Synthetic Data Backtesting Framework

• Role: ML Engineer
• Summary:
  A backtesting framework using synthetic return paths generated by statistical and GAN-based models to evaluate robustness under distribution shifts.
• Techniques: Time Series GAN, Monte Carlo Simulation, Robust Backtesting
• Keywords: Backtesting, Synthetic Data, Model Robustness

---

🧠 Research Interests

• Causal Factor Investing
• Causal Inference for Asset Allocation
• LLM-Driven Portfolio Allocation & Decision Support
• Machine Learning-Based Asset Allocation
• Alternative Data Feature Engineering
• Explainable Financial Machine Learning
• Robust Portfolio Optimization (NCO, Bayesian Covariance)
• Synthetic Data Generation for Backtesting

---

🛠 Tech Stack

Languages

• Primary : Python
• Secondary : SQL, JavaScript, Solidity
• Others : R, Swift, C/C++

Core Libraries
NumPy, pandas, scikit-learn, PyTorch, statsmodels, econml

Domains
Financial Machine Learning, Quantitative Finance, Time Series Modeling, Causal ML, Portfolio Optimization

---

📫 Contact

Email: junghun1013@icloud.com
GitHub: https://github.com/tommylee1013