Thursday returned 2 first-pick winners from 12 races, a 16.7% strike rate that sat below the model's 7-day average of 23.6%. With only 12 races in the dataset it is a thin sample to draw conclusions from, and a single extra winner would have pushed the number to 25%.
The more telling figure is the top-3 coverage. In 10 of the 12 races, the winner came from the model's three selections — an 83.3% top-3 win rate that is comfortably above the weekly average of 60.6%. The model was pointing at the right part of the field in most races; it just could not always identify the specific winner within those three.
Six first-pick dogs placed without winning, which on a place-betting strategy is a decent return. The model's first picks were not finishing fifth and sixth — they were running into the frame and finishing second and third. Those are results that keep place accumulators and each-way strategies ticking over even on a day when winners are hard to find.
Today's card is bigger, and a larger sample will give a cleaner read. Yesterday's 7-day backdrop of 836 races and a 23.6% first-pick win rate remains the more reliable guide to model health.