Carlos Ulberg needed one round. He knocked out Jiri Prochazka to win the UFC 327 main event on April 11, a fast, clean result that ended the argument early. Our model had Prochazka. We were wrong, and we will get to exactly how wrong below.\n\n### The night in brief\n\nThe card paid out in finishes. Ulberg's first-round knockout headlined a sheet where the early action did most of the talking. Cub Swanson also scored a first-round KO, and Vicente Luque closed his bout with a first-round submission. Three fights, three early stoppages, none of them dragged past the opening five minutes.\n\nThat is the honest top line: a finish-heavy night topped by a main event that ended fast. We are not going to dress the undercard with detail the record does not support. What is verified is the shape of the evening, and the shape was short.\n\n### The model autopsy\n\nThis is the part we do not hide. UFC 327 graded out at 6 of 11, or 55%. Below the line for us, and below the canon. We own it on this card.\n\nThe headliner is the cleanest miss to point at. The model picked Jiri Prochazka at 52% confidence, tagged TOSS-UP. Ulberg won by first-round knockout. So the spine of the card went against us, and at 52% the model was barely off a coin flip to begin with. That was a near-even call that landed on the wrong side, and the result, a one-round finish, gave us nothing to lean on either way.\n\nHere is the line that keeps the night from being a wipeout: the flagged picks held. We carried one LOCK-or-HIGH-confidence call into UFC 327, and it hit, 1 of 1. No LOCK miss to confess on this card. The damage was in the toss-ups and the middle of the board, where a 55% sheet does its bleeding, not in the spots where the model planted a flag and said it was sure.\n\nSo the autopsy reads two ways at once. The marquee pick was wrong, and that stings on a recap. But the part of the system built to carry weight, the flagged tier, did its job on the one chance it got. A bad card grade and a clean flagged record can live on the same night.\n\n### One honest takeaway\n\nA 55% card is a below-average night against where this model actually lives. The canonical live 2026 record sits at 67.9% across all picks, and 81.4% on the LOCK-plus-HIGH tier across 70 of those flagged calls. UFC 327 came in under both marks, and we are not going to average it away or quietly fold it into a friendlier number.\n\nThe reading that matters: a single 55% sheet is what variance looks like on a finish-heavy night topped by a 52% toss-up that resolved in one round. It does not move the canon, and it does not get to. The flagged tier going 1 of 1 here is the consistent thread, the same place the 81.4% number comes from. The all-picks miss is the cost of grading every fight on the card, toss-ups and all, in public.\n\nWe publish the misses because the alternative is a highlight reel, and a highlight reel is not a record. UFC 327 was a card we mostly got wrong, headliner included, on a night the fights ended early and a 52% call broke against us. That is the entry. The flagged picks stayed honest, the grade did not, and both go in the book.