Zian (Andy) Wang
I work across math, finance, and machine learning, and I like problems that need all three. Most of these projects started small — a class assignment, a weekend idea — but I kept pushing because the questions got more interesting the deeper I went. Now I'm looking for a graduate program where I can do more of that.
Projects
ML-Alpha — Cross-Sectional Return Prediction
Personal project — replicates and extends Gu, Kelly, and Xiu (2020)
Machine learning models for cross-sectional U.S. equity return prediction. Replicates the GKX (2020) NN5 feedforward benchmark and introduces a cross-sectional Transformer that learns stock–stock interactions via self-attention, beating NN5 with half the parameters and no hand-crafted interaction features.
Alpha Pipeline
Personal project — monthly U.S. large-cap systematic research
A monthly U.S. large-cap alpha research pipeline. Cross-sectional signal screening (IC, redundancy compression, incrementality), validation with frozen composites, and constrained mean-variance portfolio construction against a FF5+UMD factor-model covariance.
Private Credit Intelligence Engine
Evolved from an internship tool built at Ontario Teachers' Pension Plan
An agentic document intelligence system that ingests credit agreement PDFs, extracts 44 normalized covenant and pricing fields into structured data, self-verifies uncertain fields through a three-stage LLM correction loop, and generates portfolio-level risk analytics.
Causal Sentiment Engine
Personal project exploring AI-driven macro-causal reasoning
A full-stack application that models macro-financial interconnectedness as a 3D interactive causal graph. An AI agent fetches real-time data, analyzes sentiment, and propagates shocks through 52 macro nodes connected by 117 directed causal edges.
Codesight
Personal project — interactive code structure visualization
A VS Code extension that visualizes any codebase as a drillable graph — directories through files down to individual functions, classes, and types. Uses tree-sitter for static analysis with no LLM required, and exposes structural intelligence as MCP tools for Claude Code agents.
Background
Experience
Education
Mathematical Foundations
Self-directed study in graduate-level mathematics, in spare time.
| Text | Author | Coverage |
|---|---|---|
| Probability: Theory and Examples | Durrett | Ch. 1–7 — Measure-Theoretic Probability, LLN, CLT, Martingales, Markov Chains, Brownian Motion |
| Real Analysis | Folland | Ch. 0–3, Parts of 5–6 — Abstract Measures, Carathéodory Theorem, Convergence Theorems, Fubini's Theorem, Radon-Nikodym Theorem. Hilbert Spaces, Riesz Representation Theorem, Lp Spaces, Hölder and Minkowski Inequalities |
| Principles of Mathematical Analysis | Rudin | Ch. 1–7 — Metric Spaces; Basic Topology, Continuity, Differentiation, Riemann Integral, Sequences and Series of Functions; Modes of Convergence. Weierstrass Approximation |
| Introduction to Linear Algebra | Strang | Complete — Vector Spaces, Determinants, Eigenvalues and Eigenvectors, SVD, Linear Transformations |
Certificates & Continuing Study
Baruch MFE C++ Programming for Financial Engineering — Online certificate developed by Dr. Daniel Duffy and QuantNet. In progress.
Planned: Shreve Stochastic Calculus for Finance II, ODEs/PDEs, Sutton & Barto Reinforcement Learning, Raschka Build a Large Language Model (From Scratch).
Technical Skills
Italicized items indicate exposure-level familiarity; the rest I use regularly.