Coventry vs Brighton prediction: 5 of 5 AI models back Brighton

The synthesis: a hollowed-out favourite against a goalless underdog

All five experts land on the same destination — a narrow Brighton win — but the journey they describe is far more cautious than a typical favourite-versus-relegated-side script, and the caution is well-founded. The consensus probability of 47% for an away win is materially below what the opening 1.75 price implied, and below what a 5-of-5 model agreement might naively suggest. The calibrated read is that Brighton are the most likely winners, but only just, with the draw a genuinely live alternative at 29%.

Where the experts agree — and why it matters

• Coventry cannot score. Zero goals in 270 Premier League minutes, including a home blank against Hull, with Haji Wright (their most natural finisher) out long-term. Every panel member made this the backbone of the case: for Coventry to take anything, they need a clean sheet or a near-one, and they have kept exactly one clean sheet in three.
• Brighton's attack is gutted. Mitoma, Minteh, Ferguson, Wieffer, Hinshelwood, Tzimas and others are unavailable, stripping the entire first-choice wide threat and the natural centre-forward. The 8-goals-in-3 output came from a fuller squad; this is a blunted version.
• Brighton's away form is a red flag. One win in their last five league aways with 10 conceded — they score everywhere but leak everywhere, which is precisely the route to a Coventry point.

Resolving the disagreements

The sharpest split was on the scoreline: three models (Grok, GPT, GLM) called 0-1, while Claude preferred 0-2 and Kimi 1-2. The difference hinges on one question — do Coventry finally score? The 1-2 case rests on Brighton's leaky away defence giving up a goal; the 0-1 case rests on Coventry's total attacking impotence. The synthesis favours 0-1, for a simple evidential reason: a side that has failed to score against Hull's defence at home, missing its best finisher and its chief creative connectors (Eccles, Kesler-Hayden), has shown no evidence it can punish even a makeshift back line. Brighton's defensive vulnerability is real, but it requires an opponent capable of exploiting it. Coventry's underlying numbers are better than the table suggests (roughly 1.0–1.2 xG in each of the last two matches, per GPT's analysis), which keeps 1-1 and 1-2 credible — but not more likely than a clean-sheet away win.