Platform Architecture
A focused Python + FastAPI pipeline for daily momentum research.
- Universe management: liquid U.S. leaders and ETFs are refreshed into a controlled scan universe.
- Market data ingestion: OHLCV and fundamentals are collected from provider adapters and normalized into the local database.
- Rule evaluation: Minervini trend-template checks, pattern detectors, RS scoring, CANSLIM proxies, and readiness logic run over the latest bars.
- Storage + audit trail: each scan is written to persistent run/result tables so setups can be reviewed historically.
- Delivery: FastAPI renders the dashboard, export endpoints, watchlist alerts, and pre-market views from the same source of truth.
Runtime
- Backend: Python + FastAPI
- Persistence: SQLAlchemy ORM + relational database
- Jobs: scheduled scan + alert workers
- Frontend: server-rendered templates with lightweight JS enhancements
Operations
Default daily scan window: 13:00 UTC.
The design bias is operational discipline: predictable jobs, explicit fallbacks, and auditable results instead of opaque model output.