Stock Radar

01What it does

Most screeners tell you what went up. Stock Radar tries to tell you what is being accumulated before it is obvious: relative strength against the S&P 500, Weinstein stage analysis, Minervini trend templates, breakout detection, and a 0–99 RS rating computed across the whole universe rather than against an index.

02What's inside

• 586-company universe: S&P 500 plus QQQ, crypto miners, quantum, space, nuclear and metals names
• Screener with filters for score, trailing and forward P/E, street rating, analyst count and implied upside
• AI-exposure rating with tiered bonuses for infrastructure, enablers and adopters
• Capex efficiency scoring — separates companies proving revenue from companies describing it
• Weekly analyst sweep: target ranges, recommendation splits, individual actions
• Breakout detection and RS-line new highs flagged ahead of price

03How it works

Prices and fundamentals land in Postgres on a schedule; a worker scores every name and stores the components so every number in the UI can be traced back to its inputs. The AI-exposure rating is rule-based — tier overrides plus weighted components — not a model output. An LLM writes the daily market briefing and the newsletter, and it never touches a score. The API sits behind Kong for rate limiting, and the frontend is a static React build served by nginx. The test suite covers 456 backend cases.