The model is built to find races where the numbers clearly point to one horse being much stronger or weaker than the rest.
If a horse looks weak compared to its rivals but still short in the betting → it becomes a Lay bet (we bet against
it).
If a horse looks much stronger than the rest and at a fair price → it becomes a Win bet (we back it).
If the gap between horses isn’t clear, the model simply skips the race.
After a week of reflection I think the Gap 12 Lays suit email / manual betting clients far more than the 10 or 7 variants (more volume). I will stick with 12 for emails - less bets, more precise.