Charlotte +4.5 WIN: Model correctly identified the Hornets' 5-game win streak momentum and offensive surge (116.5 PPG). The value was real against an overrated Spurs spread.
DAL @ HOU UNDER 218: Injury-depleted Dallas offense performed exactly as predicted, scoring only 107 points and creating a pace/scoring environment that stayed well under.
WHAT MISSED
Houston -10.5 LOSS: Massive overconfidence in injury impact calculations. The 18.2-point adjustment (capped at 10) was wildly excessive - Dallas still had NBA-caliber players who competed hard.
Miami -5.5 LOSS: Severely underestimated Chicago's offensive capability despite their injury issues. Bulls shot lights-out (125 points) while Miami's own injury problems were underweighted.
CHI @ MIA OVER: Defense from both depleted teams was worse than expected, creating a shootout that blew past the total.
MODEL CALIBRATION NOTES
Confidence ratings were backwards: The "HIGH confidence" Houston pick was the worst miss, while medium/lower confidence picks performed better. This suggests overconfidence in injury calculations.
Systematic bias: Model dramatically overvalues star player injuries while underweighting remaining roster talent and opponent weaknesses.
METHODOLOGY ADJUSTMENTS TO CONSIDER
Cap injury adjustments at 6-8 points maximum: Current system allows unrealistic spreads. NBA games rarely exceed 15-point talent gaps regardless of injuries.
Add defensive regression factors: When both teams have significant injuries, default to higher-scoring outcomes rather than assuming one team dominates.
Overall: 2-4 record exposes fundamental flaws in injury impact calculations and overconfidence in large edges.