About

I build research-to-decision workflows that turn market and fundamental data into interpretable signals, valuation insights, and risk-aware portfolio actions — emphasizing leakage-safe evaluation, diagnostics, and practical tradability.

Resume PDF GitHub
Brownian motion paths

Featured Analytics

Research-grade deliverables with leakage-safe evaluation, interpretable diagnostics, and report-ready outputs.

Systematic Equity Alpha Research Pipeline

Factor construction, IC/t-stat screening, attribution diagnostics, and reporting designed for repeatable deployment.

IC / t-stat Validation Attribution Reporting
Report (PDF)
Alpha correlation heatmap

Monte Carlo Equity Modeling (iSoftStone)

Simulation-based scenario analysis for return distributions and tail-risk understanding.

Monte Carlo Risk Simulation Reporting
Report (PDF)
Return simulation visualization

PCA Factor Decomposition (S&P 500)

PCA on return matrices to extract latent factors, quantify explained variance, and support risk decomposition.

PCA Risk Attribution NumPy
Report (PDF)
PCA analysis visualization

Work Experience

Execution analytics, systematic research, portfolio risk modeling, and cross-functional delivery.

Ubiquant | Trade Data Analyst / Quantitative Developer

Beijing
Jul 2025 – Aug 2025
  • Extracted high-frequency trading data and reconciled internal server logs, broker submissions, and final fills to improve execution traceability, audit readiness, and data integrity
  • Identified unfilled trades across risk controls and broker routing, tracked root causes, and strengthened control diagnostics
  • Stabilized implementation shortfall at 5.4 bps by building a Python pipeline in VS Code and partnering with the HFT team, processing 30,000+ transactions per day
  • Lowered slippage variance by 17%, reducing tail-cost risk while ensuring transparency and regulatory compliance

Independent Portfolio Quantitative Researcher

Systematic Equity Strategy & Research Platform
Apr 2025 – Jul 2025
  • Developed a systematic equity trading strategy and built a SQL Server–Python research platform using the Massive Exchange API, integrating 101 Formulaic Alphas with liquidity and volume constraints
  • Filtered factors using IC analysis and regularized regressions to improve signal resilience
  • Implemented a Barra multi-factor risk model to monitor factor exposures and applied risk parity allocation to balance risk contributions
  • Modeled volatility regimes to dynamically scale capital deployment, achieving 14.4% return and 1.2 Sharpe ratio within a documented, repeatable evaluation framework
Strategy cumulative returns

North Carolina State University | Financial Modeler & Program Ambassador

Raleigh, NC
Aug 2025 – Dec 2025
  • Led a 7-member team to develop a Nasdaq equities strategy using Bloomberg Terminal and Python APIs; onboarded new members and standardized feature engineering workflows (NumPy vectorization, matrix operations)
  • Validated model reliability via hypothesis testing, leakage prevention, and time-series stability checks, confirming statistically significant out-of-sample performance across evaluation periods
  • Lowered idiosyncratic risk by 6% via diversification across a 7,000+ stock universe and improved cross-sectional sampling
  • Translated model outputs into data visualizations, presented results to industry clients, and supported program engagement

Safran Cabin, Inc. | Supply Chain Strategic Purchasing Intern

Supply Chain & Procurement
Jun 2023 – Sep 2023
  • Analyzed supply chain risk using the Newsvendor framework to evaluate demand uncertainty and service level trade-offs
  • Reduced duplicate inventory records by 20% and lowered procurement costs by 7.2% by integrating Excel (VLOOKUP, PivotTables) controls with the LeanDNA ERP system across 600+ supplier records
  • Collaborated with suppliers and cross-functional stakeholders to communicate potential disruptions, confirm order accuracy, and mitigate fulfillment risks

Quantitative Portfolio

Selected research notebooks and reports — simulation, dimensionality reduction, clustering, forecasting, and regression.

Portfolio Value at Risk Monte Carlo Simulation

January 2025 – February 2025 GitHub
  • Simulated 10,000 iSoftstone stock price paths under jump-diffusion geometric Brownian motion, analyzed period return distributions
  • Calculated 95% VaR, quantifying portfolio tail risk exposure and potential downside losses
  • Stress tested extreme volatility and jump scenarios, indicating drawdowns exceeding 50% under severe market shocks
  • Recommended protective hedging investment strategies and reduced exposure levels to mitigate extreme downside risk
View on GitHub

S&P 500 Principal Component Analysis (PCA)

October 2024 – December 2024 GitHub
  • Applied PCA to daily returns of 10 S&P 500 financial stocks with the top 3 components explaining 68.01% of variance
  • Revealed systemic risk drivers and translated results into market insights through heatmaps, component trend plots, and 3D visualization
View on GitHub

Stock K-means Classification

September 2024 – October 2024 GitHub
  • Curated Russell 2000 data (1,750 stocks) from Bloomberg and Morningstar with PEG and market-cap inputs; utilized K-means (9 clusters) to segment data-driven peer groups
  • Implemented KNN classification to tag new securities and visualized cluster outputs for interpretable equity segmentation
View on GitHub

Mortgage Rate Change Forecasting

November 2024 – December 2024 Report
  • Built a mortgage-rate change forecasting model for U.S. 30-year mortgage rates (2010–2020) using time-series and ML methods (Federal Reserve data; AR, ARMA, XGBoost Classifier)
  • Reframed forecasting into rate-direction classification, achieved a 0.67 ROC-AUC while balancing precision–recall tradeoffs
Open Report (PDF)

Iowa Housing Price Prediction

August 2024 – October 2024 GitHub
  • Developed an asset price prediction model using Ridge and Lasso regression on features such as location, floor size, and room layout
  • Achieved ~0.90 R² and ~13% out-of-sample error variance through cross-validation, indicating limited overfitting and stable performance across time windows
View on GitHub
Quant equation teaser

If it is not the end here, keep advancing forward

Back to Top