1.195
Thesis pre-OOS Sharpe
2019-2025 evaluation with parameters selected on 2010-2018 data.
Woosub ShinQuantitative Researcher
Quantitative researchbuilt to survive falsification.
I combine financial econometrics, systematic trading research, and research engineering across futures and digital assets.
My work emphasizes causal timing, out-of-sample validation, regime-dependent risk, reproducible experiments, and the separation of historical evidence from execution authority.
FuturesDigital assetsFinancial econometricsPython / SQLOOS / Walk-forwardAI-assisted development
Selected research evidence
Results stay attached to their sample, risk, and validation label.
Thesis figures come from the submitted academic paper. BTC figures describe the retained retrospective research system and are not a live or paper trading track record.
p = 0.004
EGARCH ablation
Circular block bootstrap for the filtered versus no-filter thesis comparison.
+165.92%
Retained BTC research return
Retrospective FULL result for the frozen Daily EMA 50/200 long/flat system at 5bp accounting.
-29.37%
Retained BTC MaxDD
Drawdown stays attached to the headline result instead of being hidden behind return alone.
BTC retained system: 13/13 frozen deep-validation gates passed10bp retrospective cost stress: +165.12%No live-track-record claim
Selected work
Research that shows the result and the machinery behind it.
The portfolio is concentrated around digital assets, futures, validation, financial econometrics, and reproducible research systems.
Digital asset systematic research
01BTC Research Program
A research program for BTC market states, low-turnover systematic candidates, cost stress, frozen validation, prospective observation, and explicit research-to-execution boundaries.
Open the BTC research systemMSc financial econometrics
02Volatility Regime Filtering
EGARCH-conditioned intraday NQ, ES, and WTI Crude Oil futures research with ablation, alternative volatility filters, walk-forward evaluation, placebo tests, and bootstrap inference.
Read the thesis projectResearch engineering
03Multi-Asset Research Lab
An asset-neutral framework for data contracts, causal timing, deterministic replay, provenance, reproducible experiments, frozen evidence, and auditable research infrastructure.
Explore the research platformTime-series diagnostics
04Bitcoin Bubble Detection
GSADF-based explosive-root testing for Bitcoin, separating statistical evidence from trading claims and market narratives.
Open the GSADF projectResearch profile
Evidence first. Engineering underneath.
I care about whether the information was actually available at the time, whether the experiment can be rerun, and whether the result survives attempts to falsify it.
01
Quantitative research
Financial econometrics, volatility modeling, time-series diagnostics, hypothesis testing, transaction-cost modeling, systematic strategy research, and market-regime analysis.
EGARCHGARCH/GJRGSADFBootstrapPlacebo tests
02
Validation and falsification
Out-of-sample evaluation, walk-forward testing, ablation, cost stress, subperiod analysis, causal timing checks, and explicit controls against look-ahead bias and data leakage.
OOSWalk-forwardAblationLeakage controlCost stress
03
Research engineering
Python and SQL research pipelines with deterministic replay, frozen snapshots, content hashing, experiment provenance, CI, failure-state handling, Linux, and AWS.
PythonSQLGitLinuxAWS
04
AI-augmented development
Codex and Claude are used as implementation and review tools inside explicit contracts, tests, source-control boundaries, and independent verification rather than as research authority.
CodexClaudeTest-driven reviewAudit trails
Research operating chain
Define, reproduce, falsify, then observe forward.
- Define01
Freeze the question before inspecting the answer.
Lock the dataset, information set, parameters, accounting, and failure rules before interpreting results.
- Build02
Make the experiment reproducible.
Bind source data, code, parameters, and outputs to deterministic identities that can be reconstructed.
- Falsify03
Try to break the result.
Use ablation, placebo tests, cost stress, subperiods, concentration checks, walk-forward testing, and resampling where appropriate.
- Forward04
Let new time arrive after the freeze.
Historical evidence can graduate into prospective observation, but it is not automatically execution authority.
Quant research · Systematic trading · Research engineering
Looking for teams where research quality has to survive contact with real markets.
I am interested in quantitative research, systematic trading, digital-asset research, and research-engineering roles where rigorous validation matters as much as the initial idea.
Live system telemetry
Live observation stays separate from research evidence.
Sanitized read-only Binance USD-M telemetry is refreshed every 30 seconds. Every non-zero open position sits beside flow-adjusted trading performance since August 1, 2026, while exact size, prices, balances, credentials, and execution authority stay private.
Open full telemetryFlow-adjusted trading performance
Performance Since August 1
Loading flow-adjusted performance…
All Open Binance USD-M Positions
…Loading account-wide open-position telemetry…
Read-only public projection30s pollingPerformance telemetry onlyExternal action disabled