The system spotlight this week falls on a filter that does exactly what the name says — picks the top-rated dog at Hove, but only in races that fall outside A2, A6, D3 and OR3 grades. It's a targeted exclusion list that removes the grades where the model's confidence is historically lower, and this week the result has been striking.
Three runs, one winner, one placer. An 83.3% win ROI across those three selections is obviously a small sample — nobody should be reading too much into three races — but the logic behind the system is sound. Systems that exclude specific problematic grades are often built from genuine empirical observation: someone looked at model performance by grade, found that certain class brackets produced inconsistent results, and trimmed them out. That kind of data-led filtering is exactly how sustainable systems tend to get built.
The place side has gone negative at -29.2% ROI, which with only three runs isn't statistically meaningful. What matters more is the principle: when the system decides to bet, it's been right at a rate that substantially beats the market on the win side.
To track how this system performs as the week continues, the Systems page shows the full P&L history, daily selections, and breakdown by strike rate. If the Hove Pick1 system is running live today, you'll see those picks there too. Systems with a 90-day positive ROI backdrop and a rational filtering logic are worth keeping an eye on.