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 | 1099 | 1.005 | 51% |
| Base rate | 1099 | 1.048 | 49% |
| Recent form | 1099 | 1.025 | 49% |
| Betting market | 1099 | 0.981 | 53% |
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 | 273 | 0.987 | 0.964 | −2.4% |
| Polymarket | 9 Oct 2025 | 391 | 0.994 | 0.973 | −2.1% |
| Kalshi | 1 Nov 2025 | 350 | 0.994 | 0.969 | −2.6% |
Brazil's home advantage is one of the biggest in football: long flights across a continent-sized country, heat and humidity in the north; the model's home edge is fit on the tuning seasons.
The market comparison uses the bookmakers' closing odds with their margin taken out.
Libertadores places depend on the cup winners too; "Libertadores places" here is the top six.
Clubs rest players for the Libertadores and the Copa do Brasil; the ratings only see results.
Questions and answers
How accurate are PodiumOracle's Brasileirão predictions?
On 1,099 games from held-back seasons (2024 to 2026), which the model never saw while it was tuned, the side it favoured won 51% of the time. Its log loss was 1.005, against 1.048 for the base rate and 1.025 for recent form, and 0.981 for the closing betting odds. Against the betting markets on the same finished games, its log loss was 0.987 against the bookmakers' 0.964 over 273 games, 0.994 against Polymarket's 0.973 over 391 games and 0.994 against Kalshi's 0.969 over 350 games (lower is better).