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 | 1572 | 1.046 | 47% |
| Base rate | 1572 | 1.070 | 46% |
| Recent form | 1572 | 1.061 | 46% |
| Betting market | 1572 | 1.021 | 49% |
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) | 21 Feb 2026 | 510 | 1.048 | 1.018 | −3.0% |
| Polymarket | 9 Aug 2025 | 637 | 1.037 | 1.009 | −2.8% |
| Kalshi | 28 May 2025 | 804 | 1.045 | 1.016 | −2.9% |
Home advantage is large in MLS: long journeys, time zones, altitude and artificial pitches all fall on the visiting side.
Summer heat counts as rough weather here, as well as wind and rain.
The Supporters' Shield is the 'Finish first' column.
Designated-player signings, international call-ups and midweek rotation are not in the model.
Questions and answers
How accurate are PodiumOracle's MLS predictions?
On 1,572 games from held-back seasons (2024 to 2026), which the model never saw while it was tuned, the side it favoured won 47% of the time. Its log loss was 1.046, against 1.070 for the base rate and 1.061 for recent form, and 1.021 for the closing betting odds. Against the betting markets on the same finished games, its log loss was 1.048 against the bookmakers' 1.018 over 510 games, 1.037 against Polymarket's 1.009 over 637 games and 1.045 against Kalshi's 1.016 over 804 games (lower is better).