Average Error: Mean absolute difference between predicted and actual batting average. Lower is better. <50 points = Excellent, <100 points = Good.

Brier Score & Skill Score: Measures improvement over baseline. >10% = Excellent, >5% = Good, >0% = Better than guessing.

Expected vs Actual Ratio: Shows overall calibration. 1.0 = Perfect, 0.8-1.2 = Excellent for rare events like HRs.

Calibration Analysis: The GOLD STANDARD. If we predict 25% batting average, we should see ~25% actual batting average. Small differences = well calibrated.

Risk Tiers: Shows performance across different probability ranges. Helps identify if model is better at certain prediction levels.