Seven days of data, 1,215 races, and a clear picture of where the model stands. The headline number: across the week to June 7th, the top-rated pick won 270 races from 1,215 — a first-pick win rate of 22.2%. Any of the three top picks won in 709 races (58.4%), and placed in 1,022 (84.1%).
The week had real variation day to day, and that variation is worth understanding. Sunday June 7th was the standout: 39 winners from 131 races (29.8%), on a smaller card where the model picked up well. Saturday June 6th was the highest-volume day — 253 races — returning 63 first-pick winners (24.9%), which is above the weekly average on a full card. Both days came in meaningfully above the 22.2% weekly mean.
The difficult day was Friday June 5th: 264 races and just 42 first-pick winners, a rate of 15.9%. That is a tough return, and it pulled the weekly average down. Friday cards at this time of year tend to carry a higher proportion of competitive mid-grade races where the field quality bunches up — exactly the conditions where separating dogs on composite score is hardest. The model is not blind to this; race confidence scores flag these events, but the outcomes still landed below par on the day.
Monday June 2nd produced 30 winners from 107 races (28%), which is a strong return on a mid-sized card. June 4th and June 3rd sat around 20-22% — squarely average.
The any-top-3 rate of 58.4% is the number that matters most for punters using systematic approaches. Across the week, more than half of all races were won by one of the model's top three picks. That is the base the site is built on, and it has held consistently through a mix of good and difficult days.