Le Havre vs Angers prediction: 5 of 5 AI models back Angers

The synthesis

All five experts land on the same side, and critically, they arrive there through the same mechanism rather than through imitation: Le Havre's availability crisis is the decisive asymmetry in an otherwise even-looking fixture. The panel's verification of the injury feed is consistent — eight absentees for the home side including suspended Mwanga, injured first-choice keeper Mpasi-Nzau, and a stripped centre-back pool (Zagadou, Pembélé) — against only two questionable midfielders for Angers. In a game between two of the league's weakest attacks, where fine margins decide everything, the side fielding a recognisable XI against a patched-together one holds the edge. This is not a form or table-position argument; it is a personnel argument, and it is the strongest kind because it is verifiable rather than narrative.

Where the experts disagree — and how it resolves

The meaningful split is on the scoreline shape, not the result. Claude Opus and Grok see a 0-1 grind; GLM, GPT-5.6 and Kimi see 1-2. The difference turns on whether Le Havre score at all. GPT-5.6 makes the most persuasive case for goals at both ends: this is Le Havre's first genuinely winnable-looking home fixture, so they will not sit passively — they will push their wing-backs forward to supply Maja and Samatta, which is precisely the structural weakness Angers' transition-heavy wide attackers (Allevinah, Sbaï, El Ouazzani) exploit. Against that, Le Havre's 2 goals in 3 games with their best attacking options unavailable is a real counter. On balance, the 1-2 shape is slightly more probable than 0-1 because Le Havre's home concession rate (8 in their last four home games) and Angers' leaky defence point toward an exchange of goals, while Le Havre's own scoring struggles cap the total at three. 1-2 is the modal scoreline.

The second disagreement is calibration. All five models honestly flag the same counter-case: Angers are a genuinely bad attacking side (a pitiful scoring rate in 2026 league play), they lost at home to both Lille and Rennes, and away-team favouritism between two relegation candidates is fragile. Claude Opus is the most conservative (0.40 away, 0.30 draw — nearly a coin-flip), while GLM and Kimi run it to 0.41. The panel consensus of 0.39/0.30/031 correctly sits at the cautious end, and I would not inflate it: five models clustering on the same injury-driven call is supporting evidence, not proof, and the base models' known favourite-lean means the true away probability is more likely at or slightly below 0.39 than above it.