Skip to content
View MeilinP's full-sized avatar

Block or report MeilinP

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
MeilinP/README.md

Meilin Pan

Quantitative Research | Machine Learning | Data Science

I am an applied data scientist and quantitative researcher with an M.S. in Applied Data Science from the University of Southern California and a B.S. in Statistics and Data Science from the University of California, Santa Barbara.

I currently research systematic equity signals as a WorldQuant BRAIN Research Consultant. My interests sit at the intersection of statistical modeling, machine learning, financial markets, and reproducible research.

I have engaged in independent discretionary options trading since October 2024, with a focus on volatility, option pricing, Greeks, position construction, and risk management.

Education

  • M.S. Applied Data Science, University of Southern California (2026)
    Coursework: Data Mining, Machine Learning, Database Systems, Bias in Large Language Models

  • B.S. Statistics and Data Science, University of California, Santa Barbara
    Coursework: Stochastic Processes I & II, Differential Equations, Nonparametric Methods, Experimental Design, Python and Object-Oriented Programming, Loss Models, Derivative Markets, Fixed Income Markets, Regression Analysis

Selected Work

Quantitative Research and Financial Modeling

  • Derivatives Pricing & Risk
    End-to-end chain from market quotes to a risk number: a 40-year SOFR OIS curve bootstrapped from SR3 futures and swaps under exact market conventions, a SABR calibration of the SPY implied-volatility surface with static-arbitrage diagnostics, Greek P&L attribution on an eight-leg option book with FRTB PLA testing, and a VaR/ES and stress-testing framework documented to SR 11-7. Second-order Greek terms cut unexplained P&L by 87.6%; the validation report states the framework's failures alongside its results.

  • Alpha Research Orchestrator
    Offline-first, evidence-gated research engine for WorldQuant BRAIN alpha research. A SQLite ledger records every candidate's identity, state transitions, and gate decisions, so every surviving alpha is reconstructable and auditable rather than a black box; deterministic code owns validation, routing, and limits, while an LLM can only propose. Ships a CLI and a local MCP server — no submission capability, every candidate stops at human review. Built on this pipeline: 36 equity alphas for WorldQuant IQC 2026, finishing 578th of 152,452 participants globally (top 0.4%).

  • IMC Prosperity 4 Review
    Reconstructed and evaluated trading strategies across market making, options, auctions, and multi-asset relative value using market replay, order reconciliation, and P&L attribution.

Machine Learning and Data Systems

  • Audio Event Detection and Notification System
    Group project integrating mobile audio collection, AWS EC2, SageMaker inference, and automated alerts. I was responsible for data preprocessing and model training.

  • Hybrid Recommendation System
    Built a PySpark-based rating prediction pipeline combining collaborative filtering with XGBoost and gradient-boosting models using user, business, check-in, photo, and interaction features.

  • Two-Stage Recommender, Bandit & A/B Evaluation
    Extends the recommender above into a full ranking system: CF retrieval → LambdaMART learning-to-rank → LinUCB contextual-bandit re-ranking → offline A/B evaluation. On the real 455K-rating Yelp dataset, item-CF retrieval beats a popularity baseline by 3.8x NDCG@10; the A/B harness adds CUPED variance reduction and always-valid (mSPRT) significance testing.

  • Credit Portfolio Analytics
    End-to-end consumer-credit case study: SQL portfolio analytics, a PD model (logistic regression + XGBoost, AUC 0.767) with a fair-lending disparate-impact check, decile-level risk-return pricing, and KMeans customer segmentation, delivered as a dashboard and a one-page executive summary. Declining or repricing the riskiest 40% of the book lifts modeled portfolio profit by 184%.

Highlights

  • WorldQuant International Quant Championship 2026: top 0.4% globally
  • IMC Prosperity 4: top 3.1% overall, 10th in China
  • Independent discretionary options trading experience since October 2024
  • Experience researching equity signals across 10 global regions and 50,000+ data fields
  • Experience analyzing semiconductor telemetry and test logs with Python and SQL

Tools

  • Programming: Python, SQL, R
  • Machine Learning: PyTorch, scikit-learn, XGBoost, statsmodels
  • Data: pandas, NumPy, SciPy, PySpark, Polars
  • Cloud and Systems: AWS EC2, SageMaker, Git
  • Quantitative Methods: option pricing and Greeks, volatility-surface modeling, interest-rate curve construction, market microstructure, stochastic processes, time-series analysis, Monte Carlo simulation, numerical optimization, statistical arbitrage, backtesting, and P&L attribution

Connect

@MeilinP's activity is private