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 | 1383 | 0.511 | 74% |
| Base rate | 1383 | 0.683 | 57% |
| Recent form | 1383 | 0.603 | 69% |
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) | 14 Apr 2026 | 82 | 0.497 | 0.466 | −6.8% |
| Kalshi | 6 Feb 2026 | 425 | 0.505 | 0.501 | −0.9% |
Only games ESPN carries are in the file: most Division I games in recent seasons, fewer in the early ones, so the first few seasons mainly warm the ratings up.
The title odds use an approximate 16-team tournament of the best teams by simulated wins and rating, on neutral fields. The real field is 18, with automatic bids for conference champions and play-in games; the model doesn't know conference membership, so it picks on merit.
Faceoff specialists and goalies matter a lot in lacrosse; the model only sees team results.
The season runs from February to Memorial Day; outside it the pages show how the last one finished.
Questions and answers
How accurate are PodiumOracle's NCAA men's lacrosse predictions?
On 1,383 games from held-back seasons (2024 to 2026), which the model never saw while it was tuned, the side it favoured won 74% of the time. Its log loss was 0.511, against 0.683 for the base rate and 0.603 for recent form. Against the betting markets on the same finished games, its log loss was 0.497 against the bookmakers' 0.466 over 82 games and 0.505 against Kalshi's 0.501 over 425 games (lower is better).