Norwich vs Bolton prediction: 5 of 5 AI models back Norwich

Synthesis: Norwich by one, with goals at both ends

All five experts independently converged on the same 2-1 scoreline and a home win in the 55–60% band, which is notable less for the agreement itself than for how similar the reasoning chains are. Every model identified the same two-sided asymmetry: Norwich score reliably at Carrow Road (7 goals in their last 3 home league games) while Bolton's away attack is close to inert (2 goals scored, 7 conceded, winless on the road). That asymmetry — not any vague home-favourite logic — is the spine of the prediction, and it survives calibration scrutiny because it is a matchup-specific split rather than a generic form read.

Where the models disagreed, and how I resolved it

The genuine disagreements were about shape, not outcome:

• How much the cup humiliation matters. Grok, GPT and Kimi treated the 5-0 at Manchester City as a non-event because Clement rotated nine players and the league XI is fresh. Claude added nuance — a bruised dressing room and a manager with four defeats in seven — but still concluded freshness beats morale concerns. I side with the majority here: the rotation fact is verifiable and mechanically important, while the morale angle is speculative.
• Confidence level. Claude sat at 0.55, explicitly calling the market's ~63% implied price a touch rich; GPT went highest at 0.60. Given the documented favourite-lean and overconfidence bias in these models, Claude's discipline is the right anchor. The calibrated landing zone is ~0.57 — a clear favourite, but one whose own volatility (lost 3 of 6, conceded 4 at Middlesbrough, no clean sheet in 14) makes the draw and a narrow away smash-and-grab genuinely live at roughly 4-in-10 combined.