Seven days, 827 races, and the prediction models have delivered a week that sits comfortably above what you'd expect from chance alone. The headline figures: the first pick won 24.8% of the time and placed 43.2%. At least one of the top three selections won in 58.9% of races and placed in 86.2%. Those are strong, stable numbers.
The best single day was Friday 30th May, when the first pick won 48 of 178 races (27.0%) and the top three collectively covered 59.0% of winners from a large sample. When you're assessing model performance, large-sample days like Friday are the most informative. Small cards amplify variance, making it harder to draw conclusions.
The softest day was Wednesday 28th May, where the first pick managed 23 wins from 131 races (17.6%). That's below the weekly average but not a disaster. The top three still found winners in 54.2% of races on Wednesday, which means the depth of the selections was holding up even when the top pick was cold. That's an important distinction: a tough day for the first pick is not necessarily a tough day for the model overall.
Across the full week, the second pick won 151 times and the third pick 132 times. Both are contributing meaningfully to the combined top-three win rate. When all three selections are performing consistently, that's a sign the rating system is ranking the field accurately rather than just getting lucky with a single pick.
The place rates reinforce the picture. A first-pick place rate of 43.2% means nearly half the time, the model's top selection is finishing in the first two. The combined top-three place rate of 86.2% means you're almost always getting at least one runner home in the places from the top three. For punters who use the model as a shortlist rather than a single selection, that's the number that matters most.