How good are these forecasts?
Every forecast scored on seasons the model never saw while its settings were chosen.
Held-back seasons
| Forecast | Forecasts | Log loss | Favourite won |
|---|---|---|---|
| The model | 2591 | 0.842 | 61% |
| Base rate | 2591 | 0.907 | 51% |
| Recent form | 2591 | 0.879 | 57% |
Against the betting markets
Each market's price before the start, scored on the same finished games as the model. A negative figure means the market was more accurate.
| Market | Since | Priced | Model log loss | Market log loss | Model vs market |
|---|---|---|---|---|---|
| Bookmakers (Pinnacle, DraftKings, bet365) | 24 Jan 2026 | 203 | 0.766 | 0.653 | −17.3% |
| Kalshi | 16 Jan 2026 | 96 | 0.880 | 0.779 | −13.0% |
A regular-season game can end tied after five minutes of overtime (a shootout only settles conference points), so the model forecasts a win, a tie and a loss, as in football.
The title odds use an approximate tournament: each conference's best team by conference record gets in, standing in for its tournament winner, the rest of the sixteen are the best left by simulated wins and rating, seeded in order into a fixed bracket on neutral ice. The PairWise ranking the committee uses and regional placement are not modelled.
Roster turnover between seasons is only a pull toward average; transfers, recruits and players leaving for the pros are not in the rating.
Goaltending is not in the rating.
Live scores come from ESPN's scoreboard, matched to College Hockey News's games by date and teams; a game ESPN doesn't list has no live score.
Questions and answers
How accurate are PodiumOracle's NCAA men's hockey predictions?
On 2,591 games from held-back seasons (2024–25 to 2026–27), which the model never saw while it was tuned, the side it favoured won 61% of the time. Its log loss was 0.842, against 0.907 for the base rate and 0.879 for recent form. Against the betting markets on the same finished games, its log loss was 0.766 against the bookmakers' 0.653 over 203 games and 0.880 against Kalshi's 0.779 over 96 games (lower is better).