Backtest Results - Sunday, March 1, 2026

SPREAD PICKS SCORECARD

Spread Record: 0-1 (-1.1u)

OVER/UNDER PICKS SCORECARD

O/U Record: 1-0 (+1.0u)

PARLAY RESULT

Parlay Result: LOSS (-0.5u)

DAILY SUMMARY

QUALITATIVE ANALYSIS

WHAT WORKED

• UNDER 237.5 in MEM @ IND hit cleanly - Model correctly identified that heavy injuries to both teams would slow pace and reduce scoring efficiency
• Injury impact assessment was directionally correct - Both teams were severely depleted, leading to lower-quality offensive execution as predicted

WHAT MISSED

• Massive spread miscalculation on IND +1.5 - Model predicted Pacers favored by 7+ points, but Memphis won by 19. This suggests the injury adjustment methodology is fundamentally flawed
• Overcompensated for Indiana's injuries - Despite losing Siakam and Haliburton, the Pacers were still competitive recently. Model failed to account for roster depth and coaching adjustments

MODEL CALIBRATION NOTES

• Insufficient sample to assess confidence calibration - Only one complete game analyzed with actual spread/total bets
• Injury adjustment cap appears too conservative - Memphis missing 4 rotation players should have been weighted more heavily than the 7-point maximum

METHODOLOGY ADJUSTMENTS TO CONSIDER

• Revise injury impact calculations - Current system treats all Tier 1 injuries equally at +5 points, but context matters (team depth, recent performance without player, position replaceability)
• Implement injury interaction effects - When multiple key players are out, the impact isn't additive - it compounds exponentially. Current 1.15x multiplier for two Tier 1 players is insufficient

The core issue: Model correctly identified injury-heavy games would be lower-scoring but completely botched the relative strength assessment between two depleted rosters.

CUMULATIVE SEASON RECORD

View original analysis