But it's fundamentally a questionable approach.
Assuming you fancy each one as strongly as any others, why should horse 1, 2, 3, etc on day 2 have any more (or less) chance of winning (or losing) than horses 2,3,4,5,6 on day 1 (assuming you had horse 1 a winner and stopped there).
Not knowing what the later results were doesn't alter the fact they were winners (if any were).
Unless you're backing in specific types of races (or, more to the point, your winners are coming from specific types of races because whatever selection process you're using 'works' better in them), then it doesn't make logical sense that you get early winners from a set of similarly-derived selections.
Maybe there's more/some value in thinking 'why are my earlier runners throwing up more winners', and (if you can identify it) then doubling down (or using your selection process) with that factor or those factors in mind.
***Edit Ah, Ian said the same thing more succinctly. Editorial skills!!