Sunday June 21st produced 120 races, and the model returned 39 first-pick winners -- a win rate of 32.5%. Against a random baseline of roughly 16-17% in a six-runner field, that is a meaningful margin. The place rate for first picks was 49.2%, meaning the top-rated dog ran into the first two in almost half of all races across the day.
The broader top-3 picture is the more striking number. The combined predictions called 82 winners from 120 races -- a 68.3% hit rate -- and landed a runner in the first two placings in 107 races (89.2%). Those figures reflect something more than picking individual form correctly: they suggest the model is reading race dynamics and pace profiles across a range of tracks, grades, and distances with reasonable consistency.
A first-pick win rate of 32.5% is comfortably above the rolling seven-day average of 26.2%, which puts Sunday toward the top end of recent performance rather than a structural jump. The 120-race sample is on the smaller side compared to weekday cards, which can allow variance to have more impact in either direction. But a 32.5% first-pick strike on any day is a result worth acknowledging.
Full results, race by race, are available in the historic results section.