POTE/docs/PR4_PLAN.md

# PR4: Phase 2 - Analytics Foundation

## Goal
Calculate abnormal returns and performance metrics for congressional trades.

## What We'll Build

### 1. Return Calculator (`src/pote/analytics/returns.py`)
```python
class ReturnCalculator:
    """Calculate returns for trades over various windows."""
    
    def calculate_trade_return(
        self,
        trade: Trade,
        window_days: int = 90
    ) -> dict:
        """
        Calculate return for a single trade.
        
        Returns:
            {
                'ticker': 'NVDA',
                'transaction_date': '2024-01-15',
                'window_days': 90,
                'entry_price': 495.00,
                'exit_price': 650.00,
                'return_pct': 31.3,
                'return_abs': 155.00
            }
        """
        pass
    
    def calculate_benchmark_return(
        self,
        start_date: date,
        end_date: date,
        benchmark: str = "SPY"  # S&P 500
    ) -> float:
        """Calculate benchmark return over period."""
        pass
    
    def calculate_abnormal_return(
        self,
        trade_return: float,
        benchmark_return: float
    ) -> float:
        """Return - Benchmark = Abnormal Return (alpha)."""
        return trade_return - benchmark_return
```

### 2. Performance Metrics (`src/pote/analytics/metrics.py`)
```python
class PerformanceMetrics:
    """Aggregate performance metrics by official, sector, etc."""
    
    def official_performance(
        self,
        official_id: int,
        window_days: int = 90
    ) -> dict:
        """
        Aggregate stats for an official.
        
        Returns:
            {
                'name': 'Nancy Pelosi',
                'total_trades': 50,
                'buy_trades': 35,
                'sell_trades': 15,
                'avg_return': 12.5,
                'avg_abnormal_return': 5.2,
                'win_rate': 0.68,
                'total_value': 2500000,
                'best_trade': {'ticker': 'NVDA', 'return': 85.3},
                'worst_trade': {'ticker': 'META', 'return': -15.2}
            }
        """
        pass
    
    def sector_analysis(self, window_days: int = 90) -> list:
        """Performance by sector (Tech, Healthcare, etc.)."""
        pass
    
    def timing_analysis(self) -> dict:
        """Analyze disclosure lag vs performance."""
        pass
```

### 3. Benchmark Comparisons (`src/pote/analytics/benchmarks.py`)
```python
class BenchmarkComparison:
    """Compare official performance vs market indices."""
    
    BENCHMARKS = {
        'SPY': 'S&P 500',
        'QQQ': 'NASDAQ-100',
        'DIA': 'Dow Jones',
        'IWM': 'Russell 2000'
    }
    
    def compare_to_market(
        self,
        official_id: int,
        benchmark: str = 'SPY',
        period_start: date = None
    ) -> dict:
        """
        Compare official's returns to market.
        
        Returns:
            {
                'official_return': 15.2,
                'benchmark_return': 8.5,
                'alpha': 6.7,
                'sharpe_ratio': 1.35,
                'win_rate_vs_market': 0.72
            }
        """
        pass
```

### 4. Database Schema Updates

Add `metrics_performance` table:
```sql
CREATE TABLE metrics_performance (
    id SERIAL PRIMARY KEY,
    official_id INTEGER REFERENCES officials(id),
    security_id INTEGER REFERENCES securities(id),
    trade_id INTEGER REFERENCES trades(id),
    
    -- Return metrics
    window_days INTEGER NOT NULL,
    entry_price DECIMAL(15, 2),
    exit_price DECIMAL(15, 2),
    return_pct DECIMAL(10, 4),
    return_abs DECIMAL(15, 2),
    
    -- Benchmark comparison
    benchmark_ticker VARCHAR(10),
    benchmark_return_pct DECIMAL(10, 4),
    abnormal_return_pct DECIMAL(10, 4),  -- alpha
    
    -- Calculated at
    calculated_at TIMESTAMP,
    
    INDEX(official_id, window_days),
    INDEX(security_id, window_days),
    INDEX(trade_id)
);
```

## Implementation Steps

1. **Create analytics module structure**
   ```
   src/pote/analytics/
   ├── __init__.py
   ├── returns.py       # Return calculations
   ├── metrics.py       # Aggregate metrics
   ├── benchmarks.py    # Benchmark comparisons
   └── utils.py         # Helper functions
   ```

2. **Add database migration**
   ```bash
   alembic revision -m "add_performance_metrics_table"
   ```

3. **Implement return calculator**
   - Fetch prices from database
   - Calculate returns for various windows (30, 60, 90, 180 days)
   - Handle edge cases (IPOs, delisting, missing data)

4. **Implement benchmark comparisons**
   - Fetch benchmark data (SPY, QQQ, etc.)
   - Calculate abnormal returns
   - Statistical significance tests

5. **Create calculation scripts**
   ```bash
   scripts/calculate_returns.py      # Calculate all returns
   scripts/update_metrics.py         # Update performance table
   scripts/analyze_official.py       # Analyze specific official
   ```

6. **Add tests**
   - Unit tests for calculators
   - Integration tests with sample data
   - Edge case handling

## Example Usage

```python
# Calculate returns for all trades
from pote.analytics.returns import ReturnCalculator
from pote.db import get_session

calculator = ReturnCalculator()

with next(get_session()) as session:
    trades = session.query(Trade).all()
    
    for trade in trades:
        for window in [30, 60, 90]:
            result = calculator.calculate_trade_return(trade, window)
            print(f"{trade.official.name} {trade.security.ticker}: "
                  f"{result['return_pct']:.1f}% ({window}d)")
```

```python
# Get official performance summary
from pote.analytics.metrics import PerformanceMetrics

metrics = PerformanceMetrics()
pelosi_stats = metrics.official_performance(official_id=1, window_days=90)

print(f"Average Return: {pelosi_stats['avg_return']:.1f}%")
print(f"Alpha: {pelosi_stats['avg_abnormal_return']:.1f}%")
print(f"Win Rate: {pelosi_stats['win_rate']:.1%}")
```

## Success Criteria

- ✅ Can calculate returns for any trade + window
- ✅ Can compare to S&P 500 benchmark
- ✅ Can generate official performance summaries
- ✅ All calculations tested and accurate
- ✅ Performance data stored efficiently
- ✅ Documentation complete

## Timeline

- Implementation: 2-3 hours
- Testing: 1 hour
- Documentation: 30 minutes
- **Total: ~4 hours**

## Next Steps After PR4

**PR5**: Clustering & Behavioral Analysis
**PR6**: Research Signals (follow_research, avoid_risk, watch)
**PR7**: API & Dashboard