Model Transparency & Track Record
Every CHD Scout prediction is recorded before tip-off and verified against actual game results. This page tracks the model's performance in real time — no cherry-picked results. You can view accuracy by date, confidence tier, and game type.
Predictions are grouped into confidence tiers based on the predicted margin. High-confidence picks (large predicted margin) achieve 85%+ accuracy because the talent gap is significant. True toss-up games (margin under 3) land around 55%, which is expected for games that could go either way. This tiered breakdown helps you understand where the model is most and least reliable.
FAQ
What is the model's overall accuracy?
The CHD Scout model has maintained 76%+ winner prediction accuracy across 4,000+ verified games this season, with a mean absolute error (MAE) of approximately 9.3 points. Every prediction is recorded before tip-off and verified against actual results — no retroactive changes.
Why are close games less accurate?
Games with a predicted margin under 3 points are inherently uncertain — they are essentially coin flips. The model achieves about 55% accuracy on these toss-up games, which is expected. High-confidence picks (predicted margin 12+ points) hit at 85%+ because the talent gap between teams is larger and easier to predict.
Data sourced from ESPN. Predictions by the CHD Scout model. Methodology | Accuracy Record