RollingAlpha Node
Rolling CAPM excess return above benchmark
Overview
Rolling Alpha measures the excess return generated above what CAPM predicts, calculated in rolling windows. It decomposes strategy returns into systematic risk (Beta) versus active skill (Alpha). Positive rolling alpha indicates consistent outperformance; negative alpha suggests underperformance despite taking systematic risk.
Unlike static alpha (calculated once), rolling alpha shows whether alpha generation is consistent over time. Declining rolling alpha warns of strategy decay. This metric directly feeds Information Ratio and separates luck from skill.
Formula & Calculation
Alpha = Actual - Expected return from risk taken
Annualize: Alpha_annual = Alpha_window × (252/window_size)
Parameters
| Parameter | Default | Description |
|---|---|---|
| period | 252 | Rolling window length (1 year) |
| benchmark | SPY | Market return source |
Common Use Cases
1. Skill Assessment
Rolling alpha > 0% consistently = true skill. Declining rolling alpha = strategy decay or changed market conditions.
2. Strategy Retirement
When rolling alpha turns negative, strategy is destroying value. Stop trading or redesign.
3. Rebalancing Timing
Use rolling alpha inflection points as signals to adjust factor exposures or strategy weights.
4. Manager Benchmarking
Compare fund managers: Consistent positive rolling alpha (after fees) indicates manager ability.
Advantages & Limitations
Advantages
- Separates skill from risk
- Shows performance changes
- Industry standard metric
- Directly interpretable
Limitations
- Assumes linear relationship (Beta)
- Past alpha ≠ future alpha
- Regression can be unstable
- Benchmark selection critical