NBA Over/Under Analysis - 2026-02-03

Summary

MetricValue
Total Picks6
Correct3
Accuracy50%
Skipped4
HIGH Conf2/3 correct (67%)
MEDIUM Conf1/3 correct (33%)

Results

GamePickConfidenceVegasPredictedActualResultMargin
--------------------------------------------------------------
DEN @ DETOVER 228.9MEDIUM228.9231.0245OVER+16.1
UTA @ INDSKIP-236.6-253OVER-
NYK @ WASSKIP-227.6-233OVER-
ATL @ MIAUNDER 241.1MEDIUM241.1238.0242OVER+0.9
LAL @ BKNSKIP-222.4-234OVER-
BOS @ DALUNDER 223.2HIGH223.2219.0210UNDER-13.2
CHI @ MILUNDER 223.4MEDIUM223.4220.0246OVER+22.6
ORL @ OKCSKIP-219.6-220OVER-
PHI @ GSWUNDER 221.4HIGH221.4218.0207UNDER-14.4
PHX @ PORUNDER 218.5HIGH218.5215.0255OVER+36.5

All 10 games went OVER or UNDER vs Vegas:

OVER continues to massively outperform UNDER in our system.

Skipped Games Analysis

GameVegasActualWould OVER have hit?
UTA @ IND236.6253YES (+16.4)
NYK @ WAS227.6233YES (+5.4)
LAL @ BKN222.4234YES (+11.6)
ORL @ OKC219.6220YES (+0.4)

All 4 skipped games went OVER. This was an extreme OVER night league-wide.

Cumulative Stats (Updated)

MetricBeforeAfterChange
-------------------------------------------
Total Predictions1824+6
Total Correct912+3
Overall Accuracy50%50%0%
HIGH Confidence100% (1/1)75% (3/4)-25%
MEDIUM Confidence47% (7/15)44% (8/18)-3%
OVER Accuracy67% (2/3)75% (3/4)+8%
UNDER Accuracy47% (7/15)45% (9/20)-2%

Lessons Learned

1. B2B remains the ONLY reliable UNDER primary signal. Without B2B, injury-based UNDER picks are 0/2 today and trending poorly overall. 2. Star absences WITHOUT B2B can go EITHER way. The PHX@POR game (255 total with Booker+Lillard OUT) proves replacement players can create higher-scoring chaos. 3. Dual star absences only work for UNDER when COMBINED with B2B (PHI@GSW) or extreme roster depletion on BOTH sides (BOS@DAL where DAL scored only 100). 4. OVER picks continue to outperform UNDER (75% vs 45% cumulative). The system should lean OVER more aggressively. 5. Giannis OUT = potential OVER, not UNDER. His defensive presence may suppress scoring more than his offensive production generates. His absence opened up a 246-point game. 6. MEDIUM confidence on UNDER without B2B should be downgraded to SKIP. 7. PHX@POR is the system's biggest lesson: 0% accuracy on PHX entering the day, no data on POR, yet we assigned HIGH confidence based on injuries alone. Never assign HIGH without B2B or proven team trend data.

Rule Adjustments Recommended

1. HIGH confidence UNDER now REQUIRES B2B — star absences alone are insufficient 2. Star absence adjustment cap: -3 pts max per star even for 25+ PPG players (replacement effect offsets some) 3. Consider the DEFENSIVE impact of star absences — defensive stars OUT may INCREASE total 4. OVER picks deserve more weight — 75% accuracy vs 45% for UNDER 5. Never assign HIGH on teams with 0% prediction accuracy regardless of injury factors


_Generated: 2026-02-04_ | _System Version: O/U Predictor v1.0_