Seven days of data, 1,195 races, and a story that's easier to read than most weekly model reviews.
The headline number is a 26.4% first-pick win rate across the full week. To put that in context: a random pick in a six-runner field wins about 16.7% of the time. The model's top selection wins at 26.4% — that's roughly 58% more often than you'd expect by chance. Over nearly 1,200 races, that's not noise; it's a consistent signal.
The top-3 coverage numbers are even more useful for day-to-day punting. In 64.4% of races across the week, the winner was one of the model's three rated selections. In 88.5% of races, at least one of the top three finished in the first two. If you're using the ratings as a shortlist rather than a single selection, you're working from the right pool in the vast majority of cases.
The day-by-day split reveals one weak day in the middle of the week. Sunday the 14th came in at 23.5% — below par — but that was surrounded by stronger performances on either side. June 15 and 16 both came in at 28.3% and 28.4% respectively, the two best days of the week. The model appears to be in good form heading into today.
One thing to note: there's no single model outperforming dramatically. The ratings are a composite, and the week's consistency across very different race types and venues suggests the blend is working as intended. If you want the granular breakdown by day, the Today's Results page carries the full accuracy metrics.