Backtest Results - Sunday, March 15, 2026
SPREAD PICKS SCORECARD
| Pick |
Result |
Outcome |
| Minnesota +8.5 (if Edwards plays) |
Could not match team: Minnesota +8.5 (if Edwards plays) |
NO_RESULT |
| Philadelphia +8.5 |
Trail Blazers 103 - 76ers 109 |
Covered by 14.5 |
| New York -14.5 (if Curry sits) |
Could not match team: New York -14.5 (if Curry sits) |
NO_RESULT |
Spread Record: 1-0 (+1.0u)
OVER/UNDER PICKS SCORECARD
| Pick |
Result |
Outcome |
| POR @ PHI UNDER 228.5 |
Trail Blazers 103 - 76ers 109 (Total: 212) |
Under by 16.5 |
| GSW @ NYK OVER 219.5 |
Warriors 107 - Knicks 110 (Total: 217) |
Under by 2.5 |
| -- UNDER 0 |
Game result not found |
NO_RESULT |
O/U Record: 1-1 (-0.1u)
PARLAY RESULT
| Leg |
Pick |
Outcome |
| 1 |
Philadelphia 76ers +8.5 |
UNMATCHED |
| 2 |
POR @ PHI UNDER 228.5 |
WIN |
Parlay Result: LOSS (-0.5u)
DAILY SUMMARY
| Category |
Record |
Units |
| Spreads |
1-0 |
+1.0 |
| Over/Under |
1-1 |
-0.1 |
| Parlay |
LOSS |
-0.5 |
| Daily Total |
|
+0.4 |
QUALITATIVE ANALYSIS
WHAT WORKED
- POR @ PHI Under 228.5 hit decisively - Injury analysis was spot-on. Missing Embiid/Maxey killed PHI's offense, and the model correctly identified this would suppress scoring
- Injury impact methodology showed strength - The tiered injury system properly weighted the massive impact of losing two All-Star level players
WHAT MISSED
- GSW @ NYK Over 219.5 lost narrowly - Total was 217, missing by just 2.5 points. The conditional structure (if Curry sits) wasn't properly tracked in results
- Conditional picks created execution issues - MIN/OKC and GSW/NYK picks couldn't be matched due to conditional formatting, making actual performance assessment impossible
MODEL CALIBRATION NOTES
- Confidence ratings weren't properly tested - Only one clear result with stated confidence (MEDIUM on MIN/OKC), insufficient sample
- Conditional structure caused tracking failures - 3 of 6 picks showed "NO_RESULT" due to conditional formatting that wasn't standardized
METHODOLOGY ADJUSTMENTS TO CONSIDER
- Standardize conditional pick format - Create clear "Pick A if X, Pick B if Y" structure that can be properly tracked and evaluated
- Improve injury timeline integration - The model correctly identified injury impacts but needs better real-time player status updates to convert conditional analysis into actionable picks
The core injury analysis methodology appears sound (Under 228.5 crushed), but execution and tracking systems need refinement to properly evaluate model performance.
CUMULATIVE SEASON RECORD
| Category |
Record |
Win % |
Units |
| ATS |
44-45 |
49.4% |
-7.6 |
| O/U |
49-47 |
51.0% |
|
| Parlays |
6-30 |
|
|
| Season Total |
|
|
-7.6u |
View original analysis