Thursday served up a mammoth 170-race card and the first-pick win rate came in at 15.9% — 27 winners from 170 selections, which sits below the long-run average and reflects what was a genuinely competitive day across the tracks. First picks placed 60 times, giving a place rate of 35.3%, which held up reasonably well even if the outright win column was a little lean.
The better story sits in the broader prediction accuracy. When you widen the lens to any of the top three predicted dogs, we found a winner in 100 of the 170 races — a 58.8% strike rate that shows the model is consistently identifying the right group of contenders even on days when the top pick specifically doesn't oblige. The top-three place rate hit 79.4%, with 135 of 170 races seeing at least one of our top three in the frame. Those are numbers that confirm the underlying ratings are doing their job.
Breaking it down further, second picks actually outperformed first picks on the day with 39 winners, while third picks contributed 34 — a reminder that the margins between the top-rated dogs in most races are thin and the model's second and third choices carry real value. On a card this large, variance plays its part, and a 15.9% first-pick rate across 170 races is well within normal fluctuation.
Overall it was a day where the model's depth proved more valuable than its precision at the very top. The top three collectively found the winner in nearly six out of every ten races — and that's a platform to build on heading into the weekend.