Start 4–1 KFUM Oslo: AI predicted 1–1 ✓

A stalemate born of mutual attacking inadequacy

The four experts diverge on the result — two lean home (GPT-5.6, GLM-5.2), one away (Claude Opus 5), one draw (Grok 4.6) — but they converge overwhelmingly on one trait: this will be a low-scoring, tight match decided by margins too thin to separate confidently. That convergence is more meaningful than the result split, because the disagreement on outcome is really a disagreement about which flavour of failure prevails.

Where the models agree and why it matters

Every model identifies Start's chronic inability to convert chances at home as the binding constraint. Grok 4.6 — the strongest model by recent accuracy (57% over 7 days, 53% over 30 days) — quantifies it precisely: 1.53 xG per home match converted at just 0.75 goals, with three consecutive home blanks (0-1 Fredrikstad, 0-3 Viking, 0-3 Rosenborg). GPT-5.6 frames it as a 23 xG / 16 goals season-wide underperformance and argues for regression. But seventeen matches is a large enough sample that we should question whether this match is the one where the correction arrives — particularly against a KFUM side that, even depleted, can sit in a compact block.

Resolving the outcome disagreement

Both lean on Start's home advantage and rest period, but neither resolves the fundamental problem: Start have not scored at home in 270 minutes of football. GLM's own analysis acknowledges that a 0-0 or 1-1 draw is "equally plausible," undercutting its 1-0 call.