Initial project structure: MarketScanner - Fear-to-Fortune Trading Intelligence

Features:
- FastAPI backend with stocks, news, signals, watchlist, analytics endpoints
- React frontend with TailwindCSS dark mode trading dashboard
- Celery workers for news fetching, sentiment analysis, pattern detection
- TimescaleDB schema for time-series stock data
- Docker Compose setup for all services
- OpenAI integration for sentiment analysis

.github/copilot-instructions.md

@@ -0,0 +1,36 @@
+# Copilot Instructions for MarketScanner
+
+## Project Overview
+MarketScanner is a "fear-to-fortune" trading intelligence system that identifies buying opportunities by analyzing how stocks historically respond to panic-inducing news.
+
+## Tech Stack
+- **Backend**: Python 3.12 + FastAPI + Celery + PostgreSQL/TimescaleDB + Redis
+- **Frontend**: React 18 + TypeScript + TailwindCSS + Vite
+- **Infrastructure**: Docker + Docker Compose
+
+## Key Concepts
+1. **Panic Detection**: Monitor news sentiment and price drops to identify panic events
+2. **Pattern Matching**: Match current panic against historical recovery patterns
+3. **Confidence Scoring**: Calculate buy signal confidence based on historical success rates
+4. **Real-time Monitoring**: Continuous news and price data fetching via Celery workers
+
+## Code Style
+- Python: Follow PEP 8, use type hints, async/await for database operations
+- TypeScript: Strict mode, functional components with hooks
+- Use structured logging (structlog)
+- Prefer composition over inheritance
+
+## Database
+- TimescaleDB hypertables for time-series data (stock_prices)
+- UUID primary keys
+- Soft deletes where appropriate
+
+## API Design
+- RESTful endpoints under `/api/v1/`
+- Pydantic schemas for validation
+- Pagination with skip/limit
+
+## Testing
+- pytest for Python
+- Mock external APIs in tests
+- Test critical business logic (pattern matching, confidence scoring)