Alpha Regression Analysis Results

Best R-Squared

0.0934

(Weekly)

Average Slope

-4,313,527

PnL per Alpha Unit

Total Models

Regression Analyses

Slope: Represents the expected PnL change per unit of alpha signal
R-Squared: Measures how well alpha explains PnL variance (higher is better)
Color Coding: Green (R² ≥ 0.30), Orange (R² ≥ 0.10), Red (R² < 0.10)
Outlier Removal: "No Outliers" models exclude extreme values using z-score > 3

Regression Results

Analysis Type	Slope (PnL per Alpha)	Intercept	R-Squared	Observations
Daily	-4380200.2952	2100293.3908	0.0540	199
Weekly	-13042627.6074	6305065.2416	0.0934	41
Daily (No Outliers)	-4542519.3480	2163848.6193	0.0598	196
Weekly (No Outliers)	-13042627.6074	6305065.2416	0.0934	41
Daily (Zero Intercept)	40608.5226	0.0000	-0.0011	199
Weekly (Zero Intercept)	225204.2311	0.0000	-0.0034	41
Daily (No Outliers) (Zero Intercept)	8745.0343	0.0000	-0.0004	196
Weekly (No Outliers) (Zero Intercept)	225204.2311	0.0000	-0.0034	41

Daily vs Weekly: Daily models show short-term alpha-PnL relationships, while weekly models capture medium-term effects

With vs Without Outliers: Outlier removal often improves R-squared by reducing noise from extreme market events

Best Performance: The Weekly model shows the strongest alpha-PnL relationship (R² = 0.0934)

Generated: 2025-07-11 10:00:51