AI Prediction Analysis: Weeks 1-16
Historical performance analysis to determine optimal confidence thresholds
Optimal Confidence Threshold65%
Based on 140 completed games from weeks 1-16
Games at Threshold
55
Accuracy
87.3%
Avg Confidence
74.0%
ROI
+16.8%
Performance by Confidence Threshold
Compare accuracy and profitability at different minimum confidence levels
≥50%
140 games
Accuracy
71.4%
Avg Confidence
63.9%
ROI
+7.7%
≥55%
127 games
Accuracy
74.8%
Avg Confidence
65.1%
ROI
+11.4%
≥60%
80 games
Accuracy
81.3%
Avg Confidence
70.1%
ROI
+13.3%
≥65%Optimal
55 games
Accuracy
87.3%
Avg Confidence
74.0%
ROI
+16.8%
≥70%
40 games
Accuracy
87.5%
Avg Confidence
76.8%
ROI
+13.4%
≥75%
22 games
Accuracy
86.4%
Avg Confidence
81.5%
ROI
+7.3%
≥80%
14 games
Accuracy
92.9%
Avg Confidence
84.6%
ROI
+12.4%
Moneyline Bets
Win Rate71.4%
Total Games140
ROI+7.7%
Spread Bets
Cover Rate65.7%
Total Games140
Avg Margin5.2 pts
Total (O/U) Bets
Hit Rate55.7%
Total Games140
Avg Diff9.6 pts
Analysis Methodology
Confidence Threshold: Minimum prediction confidence required to place a bet. Only games where the AI model has confidence ≥ threshold are included.
Accuracy: Percentage of correct winner predictions (moneyline) at each threshold level.
ROI (Return on Investment): Simulated profit/loss assuming $100 bets on each game at the given threshold, calculated using actual moneyline odds.
Optimal Threshold: The confidence level that produces the highest ROI across all historical games.