All projects
Russell 3000 · 3 valuation models
Stock research web app · 2026
StockKitty
Your purr-sonal stock analyst.
Solo build — full-stack + infraGitHub
StockKitty values the entire Russell 3000 the way a disciplined analyst would — every ticker is run through three DCF-style models (Free Cash Flow, Excess Profit, Residual Earnings) plus an optional Dividend Discount Model, and the results are persisted so the screener answers instantly.
It's a full-stack build I own end-to-end: a Python/FastAPI backend that batches valuations on a weekly cron, a React SPA that recomputes the same models live in the browser as you tweak assumptions, and a single-container deploy that serves both.
What it does
- Values every Russell 3000 ticker through three DCF-style models (FCF, Excess Profit, Residual Earnings) plus an optional Dividend Discount Model.
- The valuation engine is implemented twice in lockstep — Python for the batch runner, JavaScript in the browser — so users recompute intrinsic value live as they change assumptions.
- Eight named portfolio strategies (Quality-at-a-Discount, Compounders, and more), each with its own screening gates and weighting scheme, built purely from cached data.
- Per-company AI-suggested valuation assumptions via the Anthropic API.
- Production-minded infra: weekly batch cron, WAL-mode SQLite with a hot-path screener cache, two-tier JWT auth with silent refresh, and Sentry monitoring — shipped as one container on Railway.
Screens