Resume · Quantitative Research

Woosub Shin

Quantitative researcher focused on financial econometrics, systematic trading, digital assets, and reproducible research engineering.

Seoul, South KoreaPythonSQLFuturesDigital assetsFinancial econometrics

Profile

I build quantitative research systems that keep data timing, model assumptions, validation, and execution authority separate. My work spans futures volatility, digital-asset systematic research, reproducible backtesting, frozen evidence, and AI-assisted research engineering.

Selected quantitative research

MSc thesis

Volatility Regime Filtering in Futures Markets

Thesis PDF ↗
  • Studied a long-only EGARCH-conditioned intraday breakout framework across Nasdaq 100, S&P 500, and WTI Crude Oil futures using 5-minute data over 2019-2025.
  • A pre-out-of-sample specification with parameters selected on 2010-2018 data achieved a Sharpe ratio of 1.195 and a cumulative return of 3.36x on $200,000 initial capital.
  • Removing EGARCH reduced Sharpe from 1.195 to 0.382; circular block bootstrap testing supported the difference at p = 0.004. Placebo and walk-forward tests were used to probe timing dependence and robustness.

Independent quantitative research

BTC Systematic Research Program

Project evidence →
  • Built Python and SQL research infrastructure for market data, deterministic replay, frozen snapshots, causal timing, experiment provenance, validation, and reporting.
  • The retained Daily EMA 50/200 long/flat research system reports a retrospective FULL return of +165.92%, Sharpe 0.769, and MaxDD -29.37% under 5bp funding-adjusted accounting.
  • The retained candidate passed 13 frozen deep-validation gates, remained positive under 10bp cost stress, and moved into prospective forward observation on 22 Aug 2026.
  • Historical research evidence is explicitly separated from live-track-record claims and from execution authority.

Research engineering

Multi-Asset Research Lab

Platform →
  • Developed an asset-neutral framework for data contracts, point-in-time semantics, deterministic replay, immutable artifacts, content hashing, and reproducible experiments.
  • Designed explicit controls for look-ahead bias, data leakage, failure states, post-selection evidence, and prospective validation boundaries.
  • Uses Git, Linux, AWS, CI, and AI-assisted implementation workflows with independent review.

Time-series research

Bitcoin Bubble Detection with GSADF

Project →

Applied right-tailed explosive-root testing to Bitcoin price dynamics, separating statistical evidence of explosive episodes from trading claims or market narratives.

Research approach

Causal timing

Only information available at decision time belongs in the information set.

Falsification

Ablation, placebo tests, cost stress, and alternative specifications are used to attack the result.

Reproducibility

Data, code, parameters, and outputs are bound to deterministic identities and frozen evidence.

Forward boundary

Historical evidence is not treated as a live track record or automatic execution permission.