The model's selection in the top-grade race of the morning. Composite at 42 is well below the field's high of 52 (Velvet Rolo), yet predictedPosition 1 signals that the ultra-composite algorithm's speed and fieldSpeed metrics are strongly in this dog's favour. In an A1 field where the composite is more predictive than in lower grades, that override requires substantial pace evidence — and the model is apparently finding it here. Trap 5 wins 14.5% at A1 — a structural disadvantage. Average performance of 67 is competitive for this grade. The selection warrants tentative confidence: if the pace data is correct Zinedine Zedog leads from near the break; if the composite tells the bigger story, Velvet Rolo wins.