The system spotlight this week goes to a tight, track-specific filter with a strong recent run. The Doncaster Pick1 system targets the top-rated composite dog at Doncaster while excluding grades A2, B5, D2 and D3 — the conditions where the model's signal tends to be noisiest.
Over 11 runs in recent results, the system has found 6 winners and 7 placings — a 54.5% win strike rate and 63.6% place rate. Win P&L sits at +5.53 units with a 50.3% ROI. A small sample, so it deserves appropriate caution. But as a signal, it suggests the composite model's performance at Doncaster in the right grade conditions is reliable enough to be worth tracking.
The logic behind the grade exclusions is sound. A2 and B5 races tend to be competitive, tightly-graded heats where small differences in preparation matter more than ratings, and form from one runner's previous races doesn't always map cleanly onto how a race will run. D2 and D3 bring a different problem — dogs in transition between class levels where the model's form data is drawn from conditions that no longer apply. Strip those out and you're left with OR, A1, A3 and similar grades where Doncaster form tends to repeat more consistently.
At 54.5% strike rate over 11 runs, the system is performing well above chance. The ROI figure is high enough to suggest genuine edge rather than a lucky run, but 11 selections is a short history. Worth watching rather than heavily staking until the sample grows — but it's currently one of the better-performing systems in the database.