Charlotte vs DC United prediction: 5 of 5 AI models back Charlotte

Charlotte edge a coin-flip where the draw is structurally underpriced

All five models converge on a Charlotte home win, and four of five land on 2-1 as the modal scoreline. That convergence is meaningful — Charlotte's home record (W3 D2 L1, 11 goals in six), their attacking depth (Pep Biel's 10 goals, Westwood's set-piece delivery, multiple secondary scorers), and DC's missing center-back Kye Rowles all point to a side that should create enough to win. But the agreement is on a narrow plurality, not a dominant favourite: every model's confidence sits between 47–52%, and the strongest performer (Claude Opus 5, 54% recent result accuracy) is also the most cautious at 47%.

The disagreement that matters: is 1-1 the real trap?

GPT-5.6 — the weakest-performing model in this set — predicted 1-1 as the single most likely scoreline while still calling home_win as the result, arguing that Charlotte's win probability is dispersed across multiple scorelines (1-0, 2-1, 3-1) while draw probability concentrates on 1-1. This is statistically coherent, and Claude Opus 5 independently arrived at a 29% draw probability, the highest of any model. The draw case rests on a concrete structural feature: DC United have drawn 9 of 20 matches (45%) under René Weiler, including five draws in their last six road outings. That is not variance — it is a tactical signature of a low-block, counter-oriented side that shrinks matches on the road. Charlotte's own defensive profile feeds this: they have conceded in 5 of their last 6, kept just one clean sheet in that span, and famously surrendered a 3-1 lead against ten-man Toronto.

Resolving the tension

Charlotte are the better team, at home, against the ninth-placed side, with a four-game regular-season H2H winning streak. Those facts justify a home-win plurality. The gap between 48% and the market's 52% is small in absolute terms but meaningful: it means the home win is fully priced and the draw is the outcome where my estimate (28%) sits furthest above the market's fair probability (~24.6%).