But you are not the selector, you are the product of the selection. The selector is the farmer or the rancher in the case of animal husbandry, and environmental pressures in the case of Natural Selection.
One thing you seem to miss (and I'm not singling you out, it's a common misconception among those that don't actually study evolution -- even including those who argue for it) is that evolution is not an event that happens to an individual, it is an ever-ongoing process that describes populations statistically.
Let's assume that you are a breeder of screwts. If you only have one male and one female, then you are stuck with just the four (or less, if there are duplicates) alleles for skin color. But if you have 100 males and 100 females,you can choose to only let red screwts breed.
If red is a dominant trait, then there are probably parents with alleles for recessive colors. Not only is one generation not enough to eliminate the recessive alleles from the next generation, but there is a chance that there will be offspring that inherit only recessive allels and do not get red skin at all. On the other hand, while only 40% of the original herd are red-skinned, 90% of the first generation offspring have red skin. Using the same breeding strategy, we get a second-generation herd of 95% red skinned. If the percentage were a smooth funtion, it would approach 100% red as a limit, but never reach that limit, but since the percentage is calculated from discrete, whole numbers of offspring, there will be a time when they other alleles are eliminated entirely.
If it is not a breeder, but the environment that is selecting for red skin, then there is nothing but opportunity that "forces" two red-skinned screwts to breed, or "prevents" non-red screwts from breeding. There will still be the occasional blue-skin or green-skin mating. The first generation will be 75% instead of 90%. The second generation will be 80% instead of 95%. A much slower approach to the goal.
Bred breeds reach uniformity before the number of changed genes present a cross-breeding problem. Great Danes can still (in principle) be bred with chihuahuas. But "wild" breeds will often accumulate enough changes that they can't cross breed. For example, Herring Gulls cannot cross-breed with Black-backed Gulls.
At what point do we decide that Herring Gulls are a different species from Black-backed Gulls? We could say that when we reach the point where there is no chance whatsoever of cross-breeding the two. But there are two problems with that idea. First, there is no way to determine exactly what generation that degree of separation occurs, so we can only declare the sub-populations to be new species some time long after the split occurs. Plus, there is the problem of hybrids.
Horses and donkeys are so different that they are clearly labelled as different species, and yet, they can hybridize. At one time it was claimed that because the hybrid offspring were sterile (not "impotent") , it didn't matter that the hybridization was possible. But a significant number of female hinneys are not only fertile, but can concieve and bear the fetus to term. And among the Great Cats (genus Panthera) all of the female hybrids are fertile and can be re-hybridized.
In conclusion, there is never a single, one time historic "event" which produces a new species, it is a slow, generational, statistical process with no clear ending points.