Saturday June 6th was a good day. Across 253 races, the model's first pick won 63 times -- a win rate of 24.9%, which sits above the week's rolling average of 22.2% and comfortably clear of the 16.7% baseline you would expect from random selection.
Including places, the first pick landed in the first two in 117 races, a place rate of 46.2%. For anyone running place-based systems, that is a solid foundation. The combined top-three predictions covered the winner in 151 of the 253 races, a 59.7% any-top-3 hit rate -- meaning the actual winner came from our top three selections in roughly six out of ten contests.
Saturday was a high-volume card (253 races is at the larger end of typical Saturday coverage), which makes the win rate more meaningful than on a 100-race day. Performance over big samples smooths out the noise.
Not every race was predictable, and the report below covers one that most certainly was not. But as a whole, Saturday's numbers represent one of the better single-day returns we have seen recently. Full breakdown by track, grade, and prediction model is on the [historic results page](/historic-results?date=2026-06-06).
