Menu
Login Get Started
All Sports Performance
AD TEST MODE
PERFORMANCE HEADER
Mock Placement 728x90
NH

NHL Hockey

hockey • Performance Deep Dive

Trained 6d ago Predicted 4w ago
All-Time Average (76 folds)
Flat ROI
+48.1%
Hit Rate
59.7%
+7.3pp vs b/e
Recent Trend (last 2 folds · Mar 6–May 21, 2026 ) ↑ Improving
Flat ROI
+50.5%
+2.3pp vs avg
Hit Rate
64.7%
Out-of-sample Accuracy
56.7%

Avg across 76 folds

AUC-ROC
0.588

0.5 = coin flip · 1.0 = perfect

Log Loss
0.6761

Lower is better

Brier Score
0.2417

Probability calibration

76 validation folds
Nov 10, 2017 – May 21, 2026

Final Holdout Test

Held back from all training and fold evaluation — the cleanest out-of-sample signal.

No holdout data available. Re-run prediction-system run --sport nhl to generate holdout metrics.

Fold-by-Fold Metrics

Supporter+ required

Supporter access required

Unlock fold-level charts to identify training stability.

Unlock Supporter

Betting Simulation

Supporter+
Recommendation ROI
Recommendation Win Rate
Universe ROI
Universe Win Rate

Supporter access required

Unlock Supporter
Model: XGBoostModel
Calibration: platt
All-Time Average (60 folds)
Flat ROI
+29.7%
Hit Rate
49.9%
-2.5pp vs b/e
Recent Trend (last 3 folds · Mar 17–May 29, 2026 ) ↑ Improving
Flat ROI
+52.7%
+23pp vs avg
Hit Rate
57.7%
Out-of-sample Accuracy
58.5%

Avg across 60 folds

AUC-ROC
0.604

0.5 = coin flip · 1.0 = perfect

Log Loss
0.6708

Lower is better

Brier Score
0.2391

Probability calibration

60 validation folds
Nov 5, 2019 – May 29, 2026

Final Holdout Test

Held back from all training and fold evaluation — the cleanest out-of-sample signal.

No holdout data available. Re-run prediction-system run --sport nhl to generate holdout metrics.

Fold-by-Fold Metrics

Supporter+ required

Supporter access required

Unlock fold-level charts to identify training stability.

Unlock Supporter

Betting Simulation

Supporter+
Recommendation ROI
Recommendation Win Rate
Universe ROI
Universe Win Rate

Supporter access required

Unlock Supporter
Model: XGBoostModel
Calibration: platt
All-Time Average (58 folds)
Flat ROI
+23.5%
Hit Rate
64.3%
+11.9pp vs b/e
Recent Trend (last 2 folds · Mar 17–May 29, 2026 ) ↓ Declining
Flat ROI
+15.3%
-8.2pp vs avg
Hit Rate
58.5%
Out-of-sample Accuracy
57.0%

Avg across 58 folds

AUC-ROC
0.600

0.5 = coin flip · 1.0 = perfect

Log Loss
0.6742

Lower is better

Brier Score
0.2408

Probability calibration

58 validation folds
Nov 5, 2019 – May 29, 2026

Final Holdout Test

Held back from all training and fold evaluation — the cleanest out-of-sample signal.

No holdout data available. Re-run prediction-system run --sport nhl to generate holdout metrics.

Fold-by-Fold Metrics

Supporter+ required

Supporter access required

Unlock fold-level charts to identify training stability.

Unlock Supporter

Betting Simulation

Supporter+
Recommendation ROI
Recommendation Win Rate
Universe ROI
Universe Win Rate

Supporter access required

Unlock Supporter
Model: XGBoostModel
Calibration: platt
Avg RMSE
2.578

Root mean squared error

Avg MAE
2.203

Mean absolute error

R² Score
-0.0080

Explained variance

76 validation folds
Nov 10, 2017 – May 21, 2026

Fold-by-Fold Metrics

Supporter+ required

Supporter access required

Unlock fold-level charts to identify training stability.

Unlock Supporter
Model: LightGBMRegressorModel
Calibration: platt
Avg RMSE
2.294

Root mean squared error

Avg MAE
1.834

Mean absolute error

R² Score
-0.0072

Explained variance

76 validation folds
Nov 10, 2017 – May 21, 2026

Fold-by-Fold Metrics

Supporter+ required

Supporter access required

Unlock fold-level charts to identify training stability.

Unlock Supporter
Model: LightGBMRegressorModel
Calibration: platt