Wang Chongyu

Personal research system — nights and weekends

An automated equity-research desk

To learn institutional research discipline properly, I built the whole stack: role-based AI agents (analyst, PM, risk, macro, scout) write dated research memos on a schedule, a valuation engine runs DCF and comparables on every name, and market data is grounded in WRDS — Wharton's institutional research database — with cross-source integrity checks.

What it does

  • 7 role-based agent workflows produce dated, filed research memos — holding briefs, weekly PM reviews, risk reports, macro briefings, idea scouting
  • DCF + comparables valuation engine computes fair value for every covered name
  • WRDS institutional data backfill with cross-source integrity checks against market feeds
  • Lookahead-safe backtester with walk-forward out-of-sample validation and crisis stress-testing
  • Daily news and macro briefs generate automatically via CI; the terminal self-deploys on every push

Why it matters

A student teaching himself institutional research discipline by building the entire stack — data, valuation, memos, and review cadence.