KHL prediction history
See the model’s picks, correct calls and mistakes. One match counts once; pending results do not affect accuracy.
Which predictions to count
Correction picks are included in all figures below and marked individually. They were not public recommendations. Both views keep skips below 52% confidence and when data is missing.
See ROI in this view →Our main markets are game totals over/under in 60 minutes and the match winner including overtime and shootouts. Periods and overtime are additional markets under evaluation; results and odds coverage are shown separately.
Filter history · All time
Forecast coverage: 87.6% · 0 technical skips · 12 skips: below 52% or no data · 5 correction picks shown
Accuracy by game day
Last 30 days with results · before result filterWinner
121 matches · times in MoscowHow results are checked
Over means more goals than the line; Under means fewer. Over 4.5 wins with 5+ goals; Under 5.5 wins with 0–5. The model selects one of these, not both. A probability is an estimate, not a guaranteed success rate.
Total accuracy is calculated for Over 4.5 and Under 5.5, preserving the saved direction. The probability corresponds to the displayed line.
One pre-match version per game is used. The winner includes overtime and shootouts. Each total is checked against its selected line and the terms saved with that prediction. Current totals cover periods 1–3; older predictions retain their original terms, shown next to the pick.
The total pick is stored with new predictions. For older records it is reconstructed from saved pre-match probabilities using the same single-pick rule; this is not a record of previously sent recommendations. Period picks use a 1.5 line and the corresponding period score. The current winner and period selection threshold is 52%; skipped picks do not affect accuracy.