Backtest Results - Tuesday, March 17, 2026

SPREAD PICKS SCORECARD

Spread Record: 1-2 (-1.2u)

O/U Record: 0-3 (-3.3u)

Leg	Pick	Outcome
1	MIA @ CHA UNDER 234.5	LOSS
2	CLE @ MIL UNDER 226.5	LOSS

Parlay Result: LOSS (-0.5u)

MIL +11.5 WIN: Correctly identified that Giannis being questionable created uncertainty, and Bucks had enough talent to keep it close despite poor record
Injury impact assessment: Washington injuries (Beal/Kuzma out) were properly weighted, though the game went over expectations

Totals catastrophically wrong (0-3): Every under bet lost by significant margins (7.5, 12.5, 14.5 points). Model severely underestimated offensive output across the board
OKC -9.5 failure: Thunder won but didn't cover, suggesting the model overvalued their dominance against a back-to-back Orlando team that kept it competitive
Miami collapse: Heat got blown out by 30 against Charlotte, indicating the model missed Charlotte's recent defensive improvements and Miami's road struggles

Totals methodology is fundamentally flawed: 0-3 record with massive misses suggests systematic undervaluation of pace/offensive efficiency
Spread calibration mixed: 1-2 record isn't terrible, but the Miami blowout suggests injury impact on team chemistry wasn't properly weighted

Recalibrate totals formula: Current pace/efficiency inputs are clearly too conservative. Need to analyze recent scoring trends more heavily than season-long averages
Add "blowout risk" factor: When injury-depleted teams face motivated opponents, consider wider variance in outcomes rather than just point adjustments

Bottom Line: Spread analysis showed some promise, but totals methodology needs complete overhaul. 1-5 overall record is unacceptable.

Category	Record	Win %	Units
ATS	47-48	49.5%	-11.3
O/U	51-51	50.0%
Parlays	6-32
Season Total			-11.3u