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 | 170 | 0.547 | 76% |
| Base rate | 170 | 0.743 | 57% |
| Recent form | 170 | 0.635 | 69% |
The championship began in 2026, so there are no past editions to score. The track record scores the model on every test between these twelve nations instead, in the same way as every other page: tuned on older seasons, scored on the last two.
The simulated tables add bonus points the union way: four for a win, two for a draw, one for losing by seven or less, one for four tries (at each side's usual rate). The finals weekend is played on neutral ground.
It replaced the Rugby Championship and the autumn internationals; the Six Nations carries on as before.
Injuries and selection are not in the model.
Questions and answers
How accurate are PodiumOracle's Nations Championship predictions?
On 170 games from held-back seasons (2024 to 2026), which the model never saw while it was tuned, the side it favoured won 76% of the time. Its log loss was 0.547, against 0.743 for the base rate and 0.635 for recent form.