Necaxa vs Puebla prediction: 5 of 5 AI models back Necaxa

Synthesis: A narrow home lean that the market has slightly overpriced

All four experts independently land on a Necaxa win, but the agreement is thinner than it looks: the confidence band runs from 42% to 54%, and three of the four models explicitly argue that the market's ~51% implied home probability is too high. That is the signature of the favourite-leaning bias this panel tends to share — when the models cluster on a modest home call against a struggling favourite, the calibrated read should sit below the market's number, not at it. I price the home win at roughly 44%, with the draw at 29% and the away win at 27% — a genuinely competitive three-way market.

Where the experts disagree, and how I resolved it

The sharpest tension is between GPT-5.6's 54% home call and Claude Opus 5's near coin-flip read. GPT's case rests almost entirely on Puebla's attacking absences — Mustre (3 of 9 league goals) out with an ankle sprain, Gómez expected to depart for Tigres, Maestro Puch lacking rhythm. If those absences hold, Puebla's away record becomes less transferable and Necaxa's superior squad should tell. That is a real, specific argument, and it is the main reason I keep the hosts narrowly ahead. But it cuts against the form data the other three emphasize: Puebla sit 9th on 10 points versus Necaxa's 13th on 7, Necaxa are winless in five since beating Monterrey on opening weekend, and their home split (6 scored, 7 conceded in four games) describes an open, high-event venue rather than a fortress. GPT's own caveat — that Necaxa allow 18+ shots per match and haven't kept a clean sheet — concedes most of the counter-case.

The second disagreement is on the shape of the game. GPT's 1-0 scoreline depends on Puebla's replacement front line drawing a blank. Weighing the two, the base-rate evidence is stronger than the absence-driven projection — key attackers miss games all the time and teams still score — so I side with the 2-1 type game state over 1-0.