Monday produced a busy card of 115 races and the model's first-pick selections returned 26 winners — a strike rate of 22.6%. That sits just below the long-run benchmark of roughly 25% for top-pick selections, so it was a slightly below-par day on the headline figure, though not dramatically so given the natural variance across a large card.
The place column offers some context. First picks placed 47 times from 115 races, a place rate of 40.9%, which means the model's top selection was in the first two home in more than four in every ten races. When you factor in the full top-three coverage, the picture improves considerably — across all three predicted positions, 75 of the 115 races produced a winner, an any-top-3 win rate of 65.2%. That is a useful benchmark for how broadly the model is reading the form, even on a day when the single top pick ran slightly cold.
The any-top-3 place count reached 108 from 115, meaning the model had a placed runner from its three selections in nearly every race on the card. That level of consistency in the lower positions suggests the ratings are generally identifying the right horses, even when the precise finishing order at the top shuffles unpredictably — as it often does in greyhound racing.
Yesterday was a day to file under workmanlike rather than spectacular. The volume was there, the broad accuracy was solid, and the first-pick win rate will look to bounce back today.