Home backing keeps the model honest across leagues

A hit rate can flatter a forecast when home teams are doing much of the work. The competition scorecard is current as of 5 October 2026. Reconciled predictions since 13 July 2026: 455. The useful comparison asks what the model adds beyond choosing the host. A strong percentage means more when that simple habit would have done less well. In Bundesliga the model called 13 of 25 reconciled predictions correctly since 13 July 2026. That is a demanding baseline. In Bundesliga the model's accuracy is 52% since 13 July 2026. A forecast has to recognise departures from a home victory without losing the advantage supplied by a league in which the hosts often succeed. In Bundesliga the model's Brier score is 0.5191 across 25 reconciled predictions. A respectable accuracy percentage alone cannot settle whether the extra judgement is useful. In Bundesliga home sides won 15 of the 25 graded matches. Frankfurt beat Mainz 3-1 on 12 September 2026 in the Bundesliga. The result gives the home preference a concrete limit. The model gave Mainz a 57.6% chance of beating Frankfurt on 12 September 2026 (Bundesliga). A frequent outcome can be a sensible starting point without being the right answer for every fixture. The model gave Frankfurt a 19.3% chance of beating Mainz on 12 September 2026 (Bundesliga). The smaller probability still belongs in the forecast. In Serie A the model called 22 of 35 reconciled predictions correctly since 13 July 2026. Here the leading choice earned more than home backing. In Serie A the model's accuracy is 62.9% since 13 July 2026. That is a stronger reason to value the forecast than the accuracy percentage in isolation. In Serie A the model's Brier score is 0.4942 across 35 reconciled predictions. The probability assessment matters too, because a correct preference does not reveal how well the alternatives were judged. In Serie A home sides won 16 of the 35 graded matches. Cagliari beat Atalanta 2-1 on 12 September 2026 in Serie A. An encouraging competition return does not promise that its next favourite will deliver. The model gave Atalanta a 60.9% chance of beating Cagliari on 12 September 2026 (Serie A). The opposing outcome remains possible, even when the gap in the forecast is substantial. The model gave Cagliari a 16.8% chance of beating Atalanta on 12 September 2026 (Serie A). Trust should leave room for that possibility. In Champions League the model called 9 of 12 reconciled predictions correctly since 13 July 2026. The early European return is encouraging, but the workload is small. In Champions League the model's accuracy is 75% since 13 July 2026. The absence of draws here cannot justify removing the draw from a future forecast. In Champions League the model's Brier score is 0.3998 across 12 reconciled predictions. An attractive percentage needs more football behind it before it can support a lasting judgement. In Champions League home sides won 7 of the 12 graded matches. In Champions League 0 of the 12 graded matches ended in a draw. Real Betis beat Lille 3-2 on 8 September 2026 in the Champions League. The alternative to the favourite had meaningful room in the forecast. The model gave Lille a 54.5% chance of beating Real Betis on 8 September 2026 (Champions League). The focus on the leading side would have hidden that possibility. The model gave Real Betis a 21.7% chance of beating Lille on 8 September 2026 (Champions League). A good overall return still needs to be judged alongside the individual contests that went against the preference. In La Liga the model called 23 of 49 reconciled predictions correctly since 13 July 2026. The advantage over home backing is modest. In La Liga the model's accuracy is 46.9% since 13 July 2026. A useful forecast needs to retain the less convenient outcomes instead of allowing a familiar home preference to dominate the reading. In La Liga the model's Brier score is 0.5755 across 49 reconciled predictions. The draw remains part of that work. In La Liga home sides won 21 of the 49 graded matches. In La Liga 10 of the 49 graded matches ended in a draw. Athletic beat Celta 2-0 on 30 August 2026 in La Liga. The smaller possibility became the important outcome. The model gave Celta a 63% chance of beating Athletic on 30 August 2026 (La Liga). A league-wide assessment of probability quality cannot erase the alternatives attached to an individual favourite. The model gave Athletic a 15.3% chance of beating Celta on 30 August 2026 (La Liga). The next forecast still deserves a complete reading. In Premier League the model called 22 of 50 reconciled predictions correctly since 13 July 2026. The alternatives put substantial pressure on a forecast that prefers the hosts. In Premier League the model's accuracy is 44% since 13 July 2026. Choosing the home venue alone leaves a great deal of football unexplained. In Premier League the model's Brier score is 0.6635 across 50 reconciled predictions. The model adds something on that comparison, while