A unanimous call that missed
The pre-match picture pointed one way for the models. Mainz were fifth on 10 points from six games, with the stronger attacking rating of the two sides (1.1117 against Bremen’s 0.9254). Bremen were eighth on nine points, with three draws in their last five. All six predict.game entrants, opus, Kimi Cool, Gemini Thunder, GPT Striker, DeepThinker and the baseline, landed on the same scoreline: 2-1 to the home side.
Bremen won 3-2 instead. The outcome was wrong for all six, and nobody had the exact score. The Brier scores show how confident each model was in the wrong result. Gemini Thunder did least damage at 0.8138, then Kimi Cool at 0.9464, the baseline at 0.9737, DeepThinker at 0.98 and GPT Striker at 1.005. Claude Keeper was worst at 1.0248, so it leaned hardest on Mainz.
The value picks split the field. Claude Keeper, Kimi Cool, GPT Striker and DeepThinker all backed the home win at 1.73 and each lost one unit. Gemini Thunder instead took Bremen to win at 4.3, which hit for a profit of 3.3 units. The baseline made no value pick. The sample is one match, but it shows why the site grades value selections separately from raw outcome accuracy: Gemini Thunder picked the same wrong scoreline as the others and still came out on top.
Mainz: goals at both ends, no home comfort
Mainz’s attack has not been the problem. Phillip Tietz had [1] before this match, and the side had won 5-0 at Hamburg and 4-3 at Gladbach. The trouble is the defence at the Mewa Arena. Kicker had already reported that [2], and this result backed that up.
The home record now reads Frankfurt 1-3, Leverkusen 3-4 and Bremen 2-3. That is three defeats and 10 goals conceded in three home games. Away from home they have conceded only three across their three games, with two clean sheets. Mainz were also dealing with absences, including [3], which thinned their attacking options.
The head-to-head offered little warning either way. Mainz won 2-0 in Bremen in March, and the Nov 2025 fixture in Mainz finished 1-1. This was the first of the three meetings in which Bremen scored more than once.
Bremen: stable XI, another late-form win
Bremen arrived with a settled side. Daniel Thioune had [4] at Paderborn, and that continuity helped a team that had drawn at Dortmund and at home to Paderborn, both 2-2. Niclas Füllkrug, the [5], has been central to that attack, and Ludovit Reis [6] after his move.
Not everything was smooth. [7] in early October, and the club had parted with squad planner Markus Pilawa and [8] in September. None of that disrupted the results on the pitch.
Bremen have now scored in each of their last five league games, and three of those finished 2-2 or higher. The model ratings had Bremen’s defence slightly ahead of Mainz’s (1.0528 against 0.9655), but the match turned on finishing, which those ratings measure only loosely. For the models, this is a reminder that a thin attack rating does not stop a side winning an open game.
What it means
The table has shifted slightly. Mainz are seventh on 10 points from seven games (W3 D1 L3, goals 16 for, 13 against). That is the same points total as before the match, so they have slipped from fifth. Bremen are sixth on 12 points (W3 D3 L1, 15 for, 14 against), a two-point lead over their opponents.
Mainz’s form line is WLWLW at the start of the month and now ends with this defeat after the 1-0 win at Schalke. It alternates between wins and losses, and the losses have all come at home. Their next fixture is away at Elversberg on 30 October. If the pattern holds, that is a better setting for them than another home game, given they have kept two clean sheets in three away matches. The site’s forecast for that game locks 24 hours before kick-off, so no pick is published yet.
Bremen have a clear run: unbeaten in five (W2 D3 since the Köln draw, plus this win), with a home game against Hoffenheim on 31 October. The question is whether they can turn draws into wins when they host. They have scored 15 goals but conceded 14, so their games are rarely quiet, and the models will have to decide how much weight to give a side that keeps finding a result. That forecast also locks 24 hours before kick-off.
Sources
- Bundesliga.com: Phillip Tietz strike extends Mainz’s winning away run as Schalke lose again bundesliga.com ↩
- Kicker: Neuer Lack oder ein Pfeifkonzert? Mainz sucht das Auswärtsgefühl kicker.de ↩
- Kicker: Martinez rüffelt Ben Seghir: "Muss lernen, seinen Körper besser zu managen" kicker.de ↩
- Bundesliga.com: Niclas Füllkrug double rescues Werder Bremen in wild draw with Paderborn bundesliga.com ↩
- Bundesliga.com: 'Drama of leaving long gone' - Niclas Füllkrug embracing Werder Bremen return bundesliga.com ↩
- Bundesliga.com: Niclas Füllkrug strikes late as Werder Bremen snatch dramatic late 2 draw at Borussia Dortmund bundesliga.com ↩
- Kicker: Der verletzungsanfällige Lieblingsschüler: Agu fehlt Werder erneut kicker.de ↩
- Kicker: "Die beste Lösung": Werder löst Vertrag mit Kownacki auf kicker.de ↩
