Open model evaluation
KHL prediction accuracy
One latest pre-match version per game. This page evaluates model probabilities; the selected published picks and skips have their own results ledger.
Selected picks and skips →57.7%168 evaluated games
Model and baseline
Validated games
318
Majority-class baseline
50.6%
Lift, percentage points
7.1
Brier score
0.2429
The baseline always selects the most common outcome in this same sample. Lower Brier scores indicate better probabilities, including confidence in incorrect predictions.
Accuracy by market
| Market | Correct / evaluated | Accuracy |
|---|---|---|
| Winner incl. OT/SO | 97 / 168 | 57.7% |
| Over 4.5 / Under 5.5 | 100 / 176 | 56.8% |
| Period 1 · Over/Under 1.5 | 23 / 40 | 57.5% |
| Period 2 · Over/Under 1.5 | 20 / 40 | 50.0% |
| Period 3 · Over/Under 1.5 | 23 / 66 | 34.8% |
Accuracy by confidence
| Confidence | Games | Predicted probability | Actual accuracy |
|---|---|---|---|
| 70%+ | 10 | 71.7% | 50.0% |
| 60–69% | 62 | 63.9% | 69.4% |
| 55–59% | 52 | 57.7% | 46.2% |
| 50–54% | 44 | 52.6% | 56.8% |
Winner accuracy by team
| Team | Correct / evaluated | Accuracy |
|---|---|---|
| Металлург Мг | 15 / 21 | 71.4% |
| Салават Юлаев | 8 / 20 | 40.0% |
| ЦСКА | 14 / 18 | 77.8% |
| Торпедо | 13 / 18 | 72.2% |
| Авангард | 9 / 18 | 50.0% |
| Динамо Мн | 7 / 18 | 38.9% |
| Ак Барс | 8 / 17 | 47.1% |
| Локомотив | 12 / 16 | 75.0% |
| Динамо М | 11 / 16 | 68.8% |
| Адмирал | 10 / 15 | 66.7% |
| Барыс | 4 / 15 | 26.7% |
| ХК Сочи | 10 / 14 | 71.4% |
| Амур | 7 / 14 | 50.0% |
| Драконы | 7 / 14 | 50.0% |
| Спартак | 7 / 14 | 50.0% |
| Лада | 12 / 13 | 92.3% |
Prediction accuracy is measured in games involving each team; the model can also choose its opponent. Small samples are unstable.