Tuesday was a smaller card than usual — 59 races across the card — and the model returned honest, workmanlike numbers rather than anything to shout about.
The first pick landed 14 winners from 59 races, a win rate of 23.7%. That is just shy of the long-run average of around 24-25%, so no cause for concern. The place rate was more encouraging: 24 of 59 first picks finished in the top two, which is 40.7% and comfortably on the right side of what we expect.
Where the model's breadth shows up is in the top-three coverage. One of the three predicted runners won in 39 out of 59 races — 66.1%. In 52 of 59 races, at least one of the top three predictions finished in the money. That is 88.1% and reflects something useful: even on days when the top pick does not land, the model is usually pointing at the right part of the field.
Yesterday was the kind of day where the betting record would depend heavily on your approach. Flat-staking on the first pick alone would have returned slightly below par. Using a wider net across the top three would have been a comfortable day. You can dig into the full results for June 23 using the link below.