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 | 716 | 0.920 | 56% |
| Base rate | 716 | 1.061 | 44% |
| Recent form | 716 | 0.956 | 55% |
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) | 28 Jan 2026 | 198 | 0.875 | 0.863 | −1.4% |
| Polymarket | 17 Oct 2025 | 307 | 0.897 | 0.879 | −2.1% |
| Kalshi | 8 Nov 2025 | 242 | 0.876 | 0.855 | −2.4% |
A draw is its own outcome: the model has a band around an even contest in which a draw is the likeliest single result, fit on the tuning seasons.
ESPN's record starts in 2021-22, so the model is tuned on only two seasons; its numbers will firm up as seasons are added.
The squads changed enormously in 2023 when the league signed stars from Europe; the ratings caught up through results, not transfers.
There are no bookmakers' odds in the history, so there is no market comparison.
Questions and answers
How accurate are PodiumOracle's Saudi Pro League predictions?
On 716 games from held-back seasons (2024–25 to 2026–27), which the model never saw while it was tuned, the side it favoured won 56% of the time. Its log loss was 0.920, against 1.061 for the base rate and 0.956 for recent form. Against the betting markets on the same finished games, its log loss was 0.875 against the bookmakers' 0.863 over 198 games, 0.897 against Polymarket's 0.879 over 307 games and 0.876 against Kalshi's 0.855 over 242 games (lower is better).