Financial Analyst · Economic Researcher

SeanMilligan

Building quantitative research at the intersection of financial markets, macroeconomics, and international capital flows.

Python Econometrics Mandarin Portland, OR
Financial Research
Out-of-sample validation · 2026
01

About

I'm a financial analyst on FP&A and Investor Relations teams at a Fortune 500 company, with a research focus on finance and China. My work sits at the intersection of financial econometrics, macroeconomic policy, and international economics.

My academic background spans Economics and Chinese Language (University of Oregon, magna cum laude), and I hold professional certification in Standard Mandarin Chinese (ACTFL). I've lived and worked in Taipei and Hong Kong, giving me ground-level perspective on Greater China financial markets.

Recent projects have involved applications of statistical tools on dynamic, time-series relationships to determine timing strategies on moving assets with pre-established or exploratory relationships.

Currently researching the impact of Chinese currency swaps on international and domestic RMB demand. Let's chat about it!

02

Research & Projects

China Macro Research - Latest

Sanctions and Regime Shifts: Offshore RMB Analysis via spot and interbank markets

Does the February 2022 Russia sanctions shock show up in offshore renminbi pricing? Using daily Bloomberg data from 2020 to 2025 and AR(1)-GARCH(1,1) intervention models with Student-t errors, I find no significant sanctions effect in the CNY-CNH basis spread once dollar dynamics are controlled for. The apparent 2022 change in the raw series is a dollar-cycle story, not an economic demand story. A companion test of the interbank funding channel (SHIBOR-HIBOR) failed to add to the literature after I identified a denomination error in one of the source series.

Python CNH-CNYSHIBOR-HIBORTime Series
View on GitHub →
Pairs Trading v2.1

The Cost of Caution: OU Half-Life Gating and the Limits of Layering

Adds an Ornstein–Uhlenbeck half-life test as a third pre-trade condition, confirming that the spread will revert to its mean within a tradeable time horizon (40-day window) before any position is entered. The additional filter meaningfully reduced trade frequency (only 22 trades over 2.5 years) trading safety for PnL. What's notable is not any single result, but the pattern: the deviation signal outperformed Bollinger Bands in every version, under every filter regime. Robustness across conditions is a harder thing to manufacture than a single good year. This repo represents the terminus of the layered-filter research arc.

PythonPolars Cointegration OU Process
View on GitHub →
Pairs Trading v1.0

Proof of Concept: Does Volatility-of-Volatility Beat Bollinger Bands?

First implementation of a mean-reversion pairs trading strategy on the KO/PEP spread, using institutional market data and Python. Tests two signal approaches: a novel "deviation of the deviation" (volatility-of-volatility) method and a standard Bollinger Band baseline across 2024 and 2025 out-of-sample data. The deviation strategy outperformed significantly in 2025, returning approximately +12% versus -2% for Bollinger Bands, despite a later start date due to its burn-in requirement.

Python CointegrationMean Reversion
View on GitHub →
China Macro Research

The Effect of Education Across Political Boundaries

An original comparative analysis of whether education spending explains the GDP per capita divergence between Hong Kong and China from 1960–2020: two economies that started at near identical income levels and arrived at vastly different destinations. Using World Bank data and log-log OLS models with lagged variables, the analysis finds a statistically significant relationship for Hong Kong (+1.85% GDP per capita per 1% education spending) but not for China, where lagged GDP dominates. The paper addresses its own limitations directly: small N, omitted variable bias from industrial policy and institutional history, and the challenge of per capita comparisons across populations that differ by a factor of 250. A foundation for future difference-in-differences work on the same question.

Rtidyverse Linear RegressionTime Series
View on GitHub →
03

Tools & Methods

Languages & Libraries
  • Python (Polars, pandas, NumPy)
  • statsmodels / scipy
  • Matplotlib / Seaborn
  • Standard Mandarin Chinese
Data & Platforms
  • Excel (financial modeling)
  • SQL
  • Git / GitHub
Methods & Frameworks
  • Time Series Econometrics
  • Cointegration & Mean Reversion
  • Non-GAAP Financial Analysis
  • Causal Inference (IV, DiD)
04

Beyond the Terminal

📷
Photography
Drawn to street and human centered photography: the geometry of cities and shapes of people, especially in Asia.
🎵
Music
Listener and occasional player. Music as counterweight to the quantitative.
Chinese Language
ACTFL certified in Standard Mandarin. Language as both professional tool and personal obsession. Maybe French next?
Let's connect

Open to research conversations, PhD program discussions, and collaborations at the intersection of China macro and finance.