The model went 5-for-12 on this card, a 42% hit rate, with nothing flagged and a wrong call on the main event it had already graded a near coin flip. We picked Youssef Zalal at 48%, a TOSS-UP. Aljamain Sterling won by decision.\n\nNo first-round finishes registered on our card sheet for this event, so the story sits with the names rather than the finishes.\n\nOne result worth flagging: Rafa Garcia won by decision as the betting underdog, with the market giving him a 43% implied chance. That made him a slight underdog the market leaned against coming through on the scorecards.\n\n### The model autopsy\n\nThis was a poor night for the model. We went 5-for-12 on the card, a 42% hit rate. That is well below where we expect to land, and it is below a coin flip across the slate.\n\nOn the flagged side, the picture is simpler: we had no LOCK or HIGH-confidence picks on this card. The flagged record reads 0-for-0. That matters for how you read the 42%. The model never put its name behind a pick here at the confidence tier we hold ourselves to. There was no conviction call to defend and none to apologize for. The 5-of-12 is entirely the all-picks pool.\n\nThe headliner is the one to own outright. We picked Youssef Zalal. Sterling won. The model graded the fight a TOSS-UP and put Zalal at 48%, a hair under even money, so this was a fight we read as a near coin flip and called the wrong way. A 48% confidence on the losing side is a soft miss rather than a blown LOCK, but it still counts on the most-watched bout of the night, and it goes in the ledger as one.\n\nPut together, the autopsy reads honestly. It was a sub-50% card with no flagged conviction to lean on and a wrong call on a main event we had already tagged as a near-even fight. There was no flagged pick to hide behind. The model simply did not have a strong read on this slate, and the all-picks number shows it.\n\n### One honest takeaway\n\nA night like this is exactly why we separate the all-picks number from the flagged number, and why we publish both. Our canonical live 2026 record sits at 67.9% across all picks and 81.4% on LOCK and HIGH-confidence calls over 70 graded fights. This card came in at 42% with nothing flagged, which is the system telling you, in advance, that it had no edge here. The honest read is not that the model is suddenly broken. It is that the model declined to commit on this slate, and the cards where it does commit are the ones carrying the 81.4% figure.\n\nWe do not blend those numbers, and we will not start now to make a bad night look better. A 42% card with zero flagged picks is a 42% card with zero flagged picks. The value of grading ourselves out loud is that you get to see the misses with the same clarity as the hits, including a wrong call on a main event we had already labeled too close to call.