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 | 480 | 1.019 | 51% |
| Base rate | 480 | 1.069 | 45% |
| Recent form | 480 | 1.042 | 47% |
| Betting market | 479 | 0.999 | 51% |
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) | 31 Jan 2026 | 283 | 1.058 | 1.033 | −2.4% |
| Polymarket | 13 Sep 2025 | 411 | 1.030 | 1.007 | −2.4% |
| Kalshi | 8 Nov 2025 | 322 | 1.048 | 1.021 | −2.6% |
Liga MX plays two tournaments a year, the Apertura (July to December) and the Clausura (January to May), each with its own table and Liguilla; each is a season here, and ratings carry from one to the next.
Home advantage is large in Mexico — Toluca and the Mexico City clubs play at altitude — and the model's single home edge is fit on the tuning tournaments.
The market comparison uses the bookmakers' closing odds with their margin taken out.
Relegation is suspended, so there are no relegation odds.
Liguilla ties are two legs; the simulation plays each as one game at the better seed's ground.
Questions and answers
How accurate are PodiumOracle's Liga MX predictions?
On 480 games from held-back seasons (Apertura 2025 to Apertura 2026), which the model never saw while it was tuned, the side it favoured won 51% of the time. Its log loss was 1.019, against 1.069 for the base rate and 1.042 for recent form, and 0.999 for the closing betting odds. Against the betting markets on the same finished games, its log loss was 1.058 against the bookmakers' 1.033 over 283 games, 1.030 against Polymarket's 1.007 over 411 games and 1.048 against Kalshi's 1.021 over 322 games (lower is better).