A research app that scores stocks on fundamentals, technicals and sentiment — then grades its own advice and lets a community follow the best track records.
The real app runs on Streamlit + a FastAPI/Next.js stack. Here's a faithful, static walkthrough of each screen — click around.
A sell-side review of holdings — stop-losses, stretched valuations, fading momentum and profit-taking, each with an urgency score.
| Holding | Unrealized | Verdict | Urgency | Why |
|---|
| Ticker | Issuer | Portfolio % | Value | Move |
|---|
Ranked by average return on members' own logged picks. Sample members and figures.
The honest part: it grades the decisions you logged — not just the picks it liked.
If the model works, higher-scored buys should earn more. Read the trend, not one row.
| AI score bucket | Picks | Avg return |
|---|
Each name gets a 0–100 score blended from three lenses, then re-weighted by the market regime. No black box — every score opens to its reasons.
Valuation, margins, revenue growth and balance-sheet health — the business behind the ticker.
Trend, momentum and moving-average posture from six months of price history.
A single batched LLM pass over recent headlines, scored neutral when no key is set.
A shared Python core powers a Streamlit research dashboard, a FastAPI backend, and a Next.js community site — with a dual SQLite/Postgres data layer, server-side email alerts, and a test suite that runs on both databases in CI.