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 | 834 | 0.628 | 64% |
| Base rate | 834 | 0.666 | 62% |
| Recent form | 834 | 0.655 | 62% |
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) | 20 Jan 2026 | 281 | 0.629 | 0.646 | 2.6% |
| Polymarket | 10 Oct 2025 | 470 | 0.630 | 0.625 | −0.7% |
| Kalshi | 12 Nov 2025 | 404 | 0.624 | 0.625 | 0.1% |
The post-season is simulated in its current form: the play-in, best-of-five playoff series with home court to the higher seed, and a Final Four on a neutral floor. Series already under way continue from their current score.
Older seasons used other formats (a Top 16 group stage, different series lengths); their games all count for the ratings.
Live scores come from the EuroLeague's game centre; the in-game chance moves with the score and the clock only.
Rosters, injuries and domestic-league fatigue are not in the model.
Questions and answers
How accurate are PodiumOracle's EuroLeague predictions?
On 834 games from held-back seasons (2024–25 to 2026–27), which the model never saw while it was tuned, the side it favoured won 64% of the time. Its log loss was 0.628, against 0.666 for the base rate and 0.655 for recent form. Against the betting markets on the same finished games, its log loss was 0.629 against the bookmakers' 0.646 over 281 games, 0.630 against Polymarket's 0.625 over 470 games and 0.624 against Kalshi's 0.625 over 404 games (lower is better).