The system with the best strike rate over the past week is one of the more focused filters in the database: Hove Pick1, which excludes A2, A6, D3, and OR3 grades. The logic behind those exclusions reflects something real — A2 and OR3 tend to throw up unpredictable results at Hove, and D3 grade races involve beginners whose form is hard to rate with confidence.
Strip those race types out, back the model's top pick in what remains, and the results this week read as follows: 5 winners from 18 selections, a strike rate of 27.8%. Place results (top two) hit 9 from 18, exactly 50%. The win P&L sits at +1.07 units, which is a 5.9% ROI for the week — in the right direction.
The place numbers are the more nuanced part of this. A 50% place rate sounds excellent, but the ROI on place bets is -19.9% this week, meaning the average place price was not long enough to make the place market worthwhile. This is a system that finds winners, not one that finds value in the each-way market. If you are using it, win-only backing is the rational approach based on recent data.
This system has been built around a long-term 21% ROI over 90 days, which is a meaningful signal at that sample size. Weeks like this one are what that number looks like in practice — steady rather than dramatic, but consistently tilted in the right direction.
No selections are flagged for today's card, so the system is sitting out rather than forcing a pick. That kind of selectivity is a feature, not a bug. You can explore all saved systems at /systems.