Wizards +7 (HIGH confidence): Model correctly identified massive line overreaction. Portland's back-to-back fatigue and Washington's home rest advantage created the exact 11-point value gap predicted.
Under totals sweep (3/3): BPI-based pace adjustments and injury impact on offensive efficiency were spot-on. Denver/Detroit hitting 216 vs 217.5 shows precise calibration.
Injury impact weighting: Trae Young's absence hurt Washington less than expected because they're already bottom-tier, validating the diminishing returns adjustment for bad teams.
WHAT MISSED
Bucks +10.5 vs 76ers: Model severely underestimated how badly Milwaukee would perform without Giannis AND Lillard. The 17-point loss suggests the injury impact multiplier for losing multiple Tier 1 players simultaneously was too conservative.
Defensive pace assumptions: 76ers-Bucks totaled 261 despite missing stars, indicating the model overweights star player impact on pace and underweights bench unit tendencies.
MODEL CALIBRATION NOTES
Confidence ratings accurate: HIGH confidence Washington pick hit decisively, while the missed pick wasn't assigned high confidence
No systematic bias detected: Injury adjustments worked well for single-star absences but failed for multiple-star scenarios
Under totals performed perfectly, suggesting pace/efficiency modeling is well-calibrated
METHODOLOGY ADJUSTMENTS TO CONSIDER
Compound injury multiplier: When teams lose 2+ Tier 1 players, apply exponential rather than additive penalty (e.g., 1.3x the sum rather than straight addition)
Bench unit pace data: Incorporate bench-heavy lineups' historical pace when stars are out, as backups often play faster/sloppier basketball than starters